To change the build, you are looking at 1000 hours of testing.
No one can make a statement with that sort of number attached.
A decent process-driven QA program will take however hours to qualify a build as may be required.
The better the product revision maps are, the better things are compartmentalized, the more surety any system test manager will have in his regression test plan.
If they were to clear out whatever cache they missed clearing out and to reset the radio _correctly_ to CDMA auto PRL then based on the feedback for the real root issues people have suffered so far - they'd be done, in my opinion.[*]
Besides - this was A release candidate. They could easily have had others in the barn they were shaking out while looking for feedback on this one.
For all we know, we've already validated whatever they were looking for by NOT having the problem with whatever it was.
None of us were read into the program. (And if the mystically-postulated HTC lurkers are here, they're certainly not talking.)
We had no criteria.
We winged this mother, and all else is speculation.
[*]Yeah, and where unicorns live the final should also have H.264 everywhere, no fps cap, the input sensitivity timing upgraded, everything out on HDMI, H.264/MPEG4 browser support, auto-CPU-overclocking, free tethering.... and that calendar event editing changed back.