Issue Description:

remquo(16, 0.625) returns (11.0,0) instead of the expected (-0.25,2). I believe that this is due to an underlying glibc bug in 2.19: https://lists.debian.org/debian-glibc/2015/06/msg00007.html

I'm adding logic to skip the tgmath test that turned this up for all Linux/x86_64 in the short-term, but longer term we should have a way to XFAIL this only for specific versions; we should also consider either:

(a) finding some way to mark a function as busted for a target in the Platform wrappers

(b) fixing this by providing our own implementation when building for that target.