Pending reset lasts across detach and causes a reset even when disabled
If I use the dr_app* interface to launch DR, allocate a lot of memory or sthg and cause schedule_reset() to be called, then detach without taking over execution, then disable resets, and then re-attach and start executing, the first dispatch entry will go and do a reset even though they're disabled.
Action items:
- Have a release-build check for reset being enabled and bail if so in both schedule_reset() and fcache_reset_all_caches_proactively().
- Clear reset_pending in detach.