We should reject hard-to-drive speakers more often

Why?  Choosing speakers to buy / avoid based on efficiency is like choosing a new car based on a sole factor such as weight or acceleration.  I am not sure if I need to tell you it doesn't make sense.  ATC drivers/speakers are one of the most inefficient speakers (around 85dB/w/m) but, with due amplification, I am pretty sure that I do not need to tell you how good they are.

Live music needs efficient speakers to reach suitable SPL levels but for smaller spaces speaker designers can flatten the curve but the impedance does go down.

@jeffrey75 If the amplifier is behaving as a voltage source and the speaker is designed for that (and 99% are) then a flat impedance curve isn’t important, and if you look at the impedance curves of many speakers, you’ll see that most speaker designers don’t value a flat impedance curve.

Class D is going to make high sensitivity speakers obsolete.

@jon_5912 As a manufacturer of class D amplifiers I can tell you this statement isn’t correct. The advantage of higher sensitivity combined with higher impedance will always result in lower distortion from the amplifier and the speaker will have greater dynamic contrast owing to less thermal compression in the drivers, plus the speaker will be less critical of speaker cables.

Until those three problems are solved easier to drive speakers will have an advantage.

I am not sure if I need to tell you it doesn't make sense.

See above.

Class D has high power and no audible distortion.  Amplifier distortion is no longer a factor.  High power class D amps more than make up for the difference in dynamic contrast between low and high efficiency speakers.  High efficiency speakers have way worse problems than thermal compression.  

@jon_5912 wrote:

High power class D amps more than make up for the difference in dynamic contrast between low and high efficiency speakers. 

And how do they make up for that? There's only so much heat that can be dissipated in a given voice coil, not least a smaller one through typical low sensitivity. Power is power, and the less efficient receiver, unless extremely capable in power handling (which could have other, potentially detrimental effects), always ends up storing more heat, with all that entails. Thermal compression as in actually overheating the VC and causing heavy compression or sending the VC up in smoke is hardly the only, if even the main consequence following here, but rather what happens way earlier as something that has actual, audible effect. The degree to which this is pronounced, and at the (early) juncture this occurs and starts becoming a problem (referencing not least to a higher eff. scenario in which it isn't) would seem to be the more important aspect to investigate here. 

High efficiency speakers have way worse problems than thermal compression. 

I don't see how they do when properly implemented.