[KDevPlatform] bae38ce: Use pthread directly for the foreground mutex, pot

[David Nolden]

We can simply wait a few weeks and then continue this discussion. If
the deadlock doesn't reappear until then, then we can switch to a
boost::threading mutex to have something cross-platform, and report
the problem to nokia.

I've added an #ifdef Q_OS_LINUX around the pthread stuff now, so there
is zero reason to bother.

@Syron: The deadlock doesn't happen while unlocking, it happens while
trying to lock. Also, take a look at lockForegroundMutexInternal,
tryLockForegroundMutexInternal and unlockForegroundMutexInternal in
foregroundlock.cpp, and try to identify a way how it should be
possible that the number of locks and unlocks doesn't match. It
exactly tracks the number of locks (the counters are protected by the
mutex itself), and has assertions to verify everything. The mutex
isn't accessed from any other place.

Greetings, David