SUT - electrical theory and practical experience

Some observations:

I always found the specs for a cartridge confusing: 0.25mV output at 5 cm/second.  Into which load?  Open circuit?  Load=Cart impedance?  Load=recommended impedance?  Why not state the output voltage and the impedance at which it is measured?  Most recommended load impedances are specified over a large range (100Ω-1K), and they are all usually 10x the cart source impedance so the voltage divider effect will be small, but in some cases, it could have a significant impact on the output.

Matching the cart source impedance into the load would only be relevant if you are trying to transfer the maximum power from source to load which is not the goal of a phono preamp.  It would be like trying to match the load a speaker presents to the output impedance of an amplifier which can have an output impedance measured in milli-Ohms.

What you do want to match between cart and load is the conjugate match of the imaginary (reactive) part of the impedance so the cart sees a purely resistive load.  The cart has a complex source impedance which may not be a simple series or parallel capacitive or inductive component but a combination of those elements.  The complex impedance will most likely change with frequency as well, complicating the matching network calculations. 

Perhaps what is needed is an S-parameter measurement file for the cart that can be used in a linear simulator to design an appropriate conjugate matching network.  Most RF devices are spec'd this way and is the only practical way to design matching networks that work;  absent this data, it's like trying to find your way out of a maze with your hands tied behind your back and wearing a blindfold.

@rauliruegas 

My point about the specified impedance along with the output voltage was to make the necessary gain calculations easier.  In your Lyra example, using a 10 Ohm load with a 7 Ohm impedance cart will drop the output level by 4.6dB, which could be significant.  The output level will only be insensitive to the load if the load is 10x the cart impedance or higher.

My point about conjugating the imaginary (reactive) impedance of the cart is to cancel out any inductive or capacitive reactance so the cart sees a purely resistive load which will give the flatest frequency response with no peaking at high frequencies.  If the complex impedance of the cart could be measured or calculated, then the proper termination could be easily determined instead of guessing or by changing loads and listening for a difference.