Possible infinite wait in ThreadBase

[cochicde]

The particular case occurs often during testing, when a device is created and then quickly destroyed.

The problem is that CThreadBase::start does not wait for the thread to actually start before returning. That means that if the caller calls stop (usually indirectly via diableHandler or something similar), the stop function will set the thread to alive = false, and join (wait indenitely).
Because the thread didn't actually start yet, the alive = false gets overwritten when the thread actually starts, so the caller waits for a thread to end which just actuallly started and gets stuck.

I think a semaphore will do the trick to coordinate this and make sure start waits for the thread before returning.