[NativeProcessLinux] Fix potential race during thread exit

Summary:
This is the same issue as we had in D9145 for thread creation. Going through the full
ThreadDidStop/RequestResume cycle can cause a deferred notification to fire, which is not correct
when we are ignoring an event and resuming the thread. In this case it doesn't matter much since
the thread will die after that anyway, but for correctness, we should do the same thing here.
Also treating the SIGTRAP case the same way.

Test Plan: Tests continue to pass.

Reviewers: chaoren, ovyalov

Subscribers: lldb-commits

Differential Revision: http://reviews.llvm.org/D9696

llvm-svn: 237445