[clang] Thread Safety Analysis: Honor try-lock branches of a void conditional (PR #211973)

[Jameson Nash via cfe-commits]

vtjnash wrote:

That appears to be a true-positive, correctly no longer suppressed by this PR, since it acquires the lock, then forgets whether or not that it did and forgets to unlock it. Did you over-reduce that MWE?

https://github.com/llvm/llvm-project/pull/211973