to buy or not to buy (ultrasonic record cleaner)

That would do certain things other brushes wouldn't. I could add one to the repertoire to try it. 

With respect,
I think it is impossible to see at that magnification what is going on; as to ultrasonic "getting into the grooves," we are really dealing with far smaller microscopic levels.
The traditional factors include transducer power, placement, bath water size, frequency and rotational speed or the amount of time exposed.
As to using a brush, I’ve been "cross-cleaning" for years, using various US machines in combo with traditional manual cleaning and using a point nozzle vacuum machine to remove fluid and/or rinse water.
There’s also the question of surfactants and whether they are effectively removed.
If the question is whether ultrasonic works, I’d say yes, having used both the commercially available machines (before the Degritter): the AD and the KL. As to how to buy, it depends on whether you want one step convenience or are willing to devote more time to the process but perhaps improve your results. You can also save money doing DIY, but that’s only one consideration, and not always the case if you go industrial or medical grade.
I will go the industrial DIY route after the KL dies, not to save $, but to benefit from a more robust feature set while maintaining quality of equipment (Some of the cheap machines you buy online burn out and are disposable).
There’s a lot more than this, but most of the science isn’t applied to vinyl, since US has been used in industry for quite a while for other things. Neil Antin has been working on an update to his original paper on Aqueous Cleaning which should soon be available for free. It will contain some interesting insights into the ultrasonic process, and addresses filtering and pumps as well as the transducers, placement, etc. as applied to the vinyl LP.
Bill Hart

There is no one best way to do this, and as to the OP, the idea that ultrasonic is going to show its stuff on a "dirty" record is, I think you’ll find, not necessarily true. In many cases with used records we have no idea what contaminants they have been exposed to; ultrasonic is one tool in your arsenal to deploy. I combine cleaning methods, pre-clean used records (some of which are rare and of high value) and use ultrasonic, but the latter is not the only solution-- it may be good for records that have been maintained in pristine condition by audiophiles, but even in those instances, I often wonder what "cleaning" methods were used that left residue.
There are best practices to this starting with what archives do, including the Library of Congress.
There is no one wonder chemical and dealing with ultrasonic, there are trade offs-- you want to lower surface tension of the water, perhaps have some detergent to bind the contaminants, but the other side of the equation is how that stuff is removed once it has done its job. That, to me, is where the rinse step plays a role.
I’m not a materials scientist or bio chemist but have spent a fair amount of time digging down into this stuff and can defer to those with more knowledge than me. Unfortunately, most of the science of making records, at least in the published papers of journals like the AES, stopped in the ’70s. Cleaning agents are regarded as proprietary by their makers and in some cases, the manufacturer of an ultrasonic machine may insist that you use their fluid lest you violate their warranty.
If you are DIY, you have more options and there are a lot of fluid concoctions that people have shared on the web. Often, it is based on trial and error, or someone’s suggestion that a particular chemical or combination of chemicals works for them.
There is science to this, and it requires a methodical approach. As mentioned, Neil Antin has done an extensive study on record cleaning, the chemistry and methods. He is in the process of revising that study to account for ultrasonic cleaning and I hope to publish it soon. It will be available at no charge, and Neil has been generous with his time in working with users to help them address issues of detergency, foaming, removal of surfactant and the like.
As to investing in an ultrasonic machine as a one time proposition, the transducers will eventually burn out; there are equipment failures (witness the early AD teething problems which I gather have been sorted with the PRO model).
If one is looking for a simple, drop a record in and push a button type approach, the current in vogue machine is the DeGritter, with which most users report positive experience. I think you’ll find if you spend time with the made for LP machines that they are a compromise; thus, the need to use conventional manual cleaning for used records, in combination with ultrasonic. There is no one size fits all solution to this in my estimation.

There's an extensive thread on the Hoffman board about the Degritter with a few users trying different approaches. Might be worth a look. The Degritter seems to be the best alternative right now in the made for LP one button approach. 
I use a big Monks and the KL, as mentioned. When the KL dies, I'll go full industrial. Have guys wearing orange hazard vests and those white suits, flashing blue warning lights, etc. 
Neil got me to use a UV light. It does show some stuff that you can't see by naked eye, but the visual really isn't that instructive. You can see plate out issues in the vinyl though. Some stuff is very consistent, other stuff is rather frightening. Unfortunately we have little to no info on vinyl compounding and even less control over it-- you buy what's made. While I don't necessarily like the sound of the old MoFi, that JVC super vinyl is still tops in my book. But,  a lot of the records I'm chasing were pressed at a nadir in vinyl quality in the U.S. early-mid 70s so-called spiritual or avant-garde jazz. 

You can manually clean pretty effectively without spending a lot of money. One of the issues isn't just whether the chemistry is effective, but how easy it is to remove from the record without leaving a residue that includes the chemistry plus contaminants.
Rinsing is good. But dishwashing detergent is pretty heavy. Dawn was popular with car detailers to strip off wax build up before they would get to work on a painted or clear coat surface. It has a lot of stuff in it that you don't need for cleaning records. Cheap yes. I think you can do better without spending much more.  

Here's the 2nd Edition of Neil Antin's Precision Aqueous Cleaning of Vinyl Records. It expands on the chemistry and processes, and includes a much expanded section on ultrasonic chemistry and processes. Not a light read, but an excellent reference. It is a free download. Neil did all the work; I just published it. 
https://thevinylpress.com/precision-aqueous-cleaning-of-vinyl-records-2nd-edition/

I preclean any used album and some new ones. Ultrasonic is good at the fine cleaning but the basic stuff I do manually via a Monks Omni. At the same time, I inspect the record-- Neil got me onto using a UV light which is instructive, but not all telling. What you can see is some particulate and differences that aren’t as visible in normal light in the vinyl itself- not that there is anything you can do about the latter.
Given how the Monks works, if a record has a warp, you will know it. In those cases, after a preclean, the record will go into an Orb made DF-2. Final cleaning via US (currently using the KL which is hanging in there). I rinse, using high grade water, a vac dry on the Monks. My set up is shown in my system photos on Audiogon, but I’ll include a shot here: https://thevinylpress.com/app/uploads/2017/12/DSCF0766.jpg

’Twas the original "automatic" record cleaning machine borne out of Percy Wilson’s papers on vinyl record contamination. There was a test mule, photograph of same floating around on the web somewhere, and Monks was the one to commercialize it. I saw my first one circa early '70s at the high end salon in Pittsburgh, run by the debonaire Tasso Spanos (lovely guy with ears and a real sense of the high end back the day). Tasso was one of the few I knew who had one-- they were really aimed at libraries, radio stations, and in the UK, facilities that would offer record cleaning as a service.
Tasso’s old machine showed up on E-Bay or Audiogon several years ago as a rusted out hulk.
When Monks the Elder died, the company became dormant, though there were still folks out there servicing them. Monks the Younger rebooted the company a decade or more ago, and this is one of the newer models-- finish may be a little nicer than the early ones, but the unit is in many respect the same British engineering- quirky, but it gets the job done and I like it as an artifact. (The pump actuator is from a Mini with a windshield wiper icon-- I don’t use the fluid brush head dispenser because I use various fluids and a pure water rinse). The machine is still made as far as I know, although Monks recently introduced a line that doesn’t rely on the thread buffer that was part of the design of these-- I have not seen or played with these newer ones, which I gather are priced at a few grand as opposed to several few grands. It’s a keeper.

I didn’t like the sonic signature of the Monks fluid. I use a Hannl fluid which is no longer made and will eventually have to find another fluid. I also use AIVS #15 for a deeper clean, agitating with a unidirectional pad applicator that came with Lloyd Walker’s Prelude kit (you can buy replacement pads, but unfortunately, not the applicator, separately). I do use the Monks brushes for the Hannl fluid and for the rinse stage. I think the rinse is critical.
After pre-cleaning, I use the KL ultrasonic and if I’m really concerned about the results, will dry via the Monks using a little more high grade water.
I think the combination of methods- essentially manual cleaning, combined with vacuum, and ultrasonic, with appropriate rinse steps, is synergistic. I’m agnostic as to product brands-- I’m looking for results. I’ve gone through a lot of records and have a pretty good regimen that I’m satisfied with, but am always open to learning.
I’m also not about ritual or voodoo-- if you can get a record clean, and remove the contaminants including those bound up with the cleaning fluid so you are not leaving a residue on the record, you should be good-- even without spending a lot of money on fancy equipment. The equipment is a convenience. I like the Monks and it is one machine I’ll keep. I like the KL too for that matter, but most ultrasonics eventually die or need parts that aren’t readily obtainable. DIY ultrasonic makes a lot of sense. For those folks that don’t want the trouble, the Degritter is apparently the current fav and I’ve heard almost uniformly positive comments about it. As mentioned, I’ll go industrial when the KL finally dies (not that I’m wishing for that, it has served me well).

Try the Monks brushes- they aren't goats hair but I have those and prefer the Monks for some things, unidirectional applicator for other things. Some of it comes down to how it applies and what you are trying to accomplish with the applicator. I've tried quite a few and everyone seems to have a preference. 
You might want to read some of Neil Antin's work on filtering. There's a whole lot to that area that was unknown to me starting with the specs and what an "absolute" rating means in terms of granularity. (I'm not a materials science so I'm sure there's a better way to describe that).