This circuit looks to be noting more than M1/OA1 acting as a constant I source,
and OA2/R1 acting as a high side current sense. Seems legit, but redundant in
that OA2 is really not needed. But it is clever in that this is a current source, not
sink, so simplified version (one OPAmp) has error term related to supply stability.
This eliminates that as long as diff amp is well balanced for R ratios that control
CM rejection.
Regarding using OAs in a fdbk loop, you have to look at phase margin of the loop.
You can calc that or use time domain techniques (disturb loop with impulse and use
scope to look at ringing/phase margin) to get an idea of what loop is doing. There is
a ton of phase margin/loop compensation ap notes online discussing these issues.
http://search.edaboard.com/measure-phase-margin-and.htmlRegards, Dana.