[clang] [analyzer] Fix null-pointer dereference in PthreadLockChecker (PR #210912)

[Balázs Benics via cfe-commits]

https://github.com/steakhal commented:

It took me a while to think this through but I think it is correct - or if not, then it's still better than what we had.

You should consider my comments across these hunks and while the comments were useful for understanding and verifying the claims, I think we could drop 80% of them to keep the most pressing 20% that actually matters.

I'd like to see some empirical evidence that this wouldn't cause FPs. I know it's probably a lot more work than the originally proposed patch, so I understand that this turned out to be a time sink for you, but this is a core component and we must get this right.

https://github.com/llvm/llvm-project/pull/210912