gthread: use inline keyword for _get_impl() functions

Give compiler a hint that these should be inlined,
which doesn't seem to happen by default with -O2.
Yields 5% speedup in artificial benchmarks, and
1% speedup in a real-world test case doing a lot
of mutex locking and unlocking.

https://bugzilla.gnome.org/show_bug.cgi?id=730807