thread safety may be broken for OwnThread operations if sequential activity used about rtt

[meyerj]

Yes, I see the problem. This affects the toolchain-2.9 branch only due to the patch introduced in f5e1d78.

I remember that while working on the patch we had a variant where RTT::base::ActivityInterface::thread() was allowed to return a null pointer for all activities where the executing thread cannot be known by design, like for SequentialActivity and SlaveActivity without a master. isSend() would always return true in that case. I am not sure anymore why this approach has been disregarded at the end. Returning os::MainThread in these cases is just a guess, which might be valid for unit tests or if the operation is called from the TaskBrowser, but not in most other applications.