[clang] [analyzer] Improve messaging in security.VAList (PR #157846)

[Donát Nagy via cfe-commits]

NagyDonat wrote:

@steakhal I looked through the code of this checker and added a few tests to ensure that all message kinds are covered by the testing (which implies high coverage for the whole checker: most of the logic is "behind" one kind of message). I was a bit surprised to see that I didn't have to add too much: it turns out that my patch preserved many existing messages and didn't introduce too many new situations.

https://github.com/llvm/llvm-project/pull/157846