Unigen 2.5.2
Release with newer integrated Arjun. This fixes some performance regressions. Currently, some of the "empty occurrence" setup is not well thought-through for UniGen, so it's partially disabled. In particular, the embedded ApproxMC doesn't make use of them. This is a shame, but we need to figure out how to do it right.
Meant to be used with:
c ApproxMC SHA revision f04ef377b41afa517eedeaa2abff40c940d4cab8c Arjun SHA revision 2a230739582afd51c264e31a7d5b21337b02dcf1c CMS SHA revision 9b4ee6f14849f7ea2b9efe2f33b698a3ba203947