v1.0-beta.7

This revision contains the following changes:

1. Some functions have been optimized to reduce code size, which unfortunately does not always work due to a GCC bug. Clang is not suffering from this issue.
2. __gthread_join() and __gthread_detach() now detect non-joinable threads and fail accordingly, instead of decrementing the reference count blindly and causing undefined behavior. This is an ABI break if your program is linked against mcfgthread without optimization enabled. Upgrading is suggested, as soon as possible.

Full Changelog: v1.0-beta.6...v1.0-beta.7