Changes to our tests caused us to start hitting a random failure of clang compiler querying due to a 10 second timeout we added to avoid another bug in which our process would randomly get permanently stuck reading stdout for clang-tidy (clang appears to have the same issue). When this timeout is hit, we should be retrying the command instead of just failing. (UPDATE: We fixed the root cause of the timeout).

So far, I've only seen this with clang and on Windows; however, I'm pretty certain this will also cause clang-tidy to fail, since tests were hitting that previously, but the failure would only be noticed if clang-tidy was expected to give warnings but did not.

The failed compiler querying can result in us falling back to use a lower C++ standard or other subsequent failures (i.e. clang-tidy can report incorrect errors or be missing warnings).