[clang] [clang][modules] Minor improvements to diagnosing `out of date` errors (PR #136612)

[Cyndy Ishida via cfe-commits]

cyndyishida wrote:

> Does the described test case look like a build system problem? In my understanding at step 4 we should have re-scanned B to know that we need some B'.pcm depending on A-2.pcm. Regardless of the answer I don't want to block the change as it is still an improvement. Just thinking if end users are even supposed to see the new diagnostic. And if it should be actionable for the end users.


It could be a build system bug or a dependency scanning bug (depending on whether you consider that a build system). The most recent case I investigated is a dependency scanning bug, but where a re-scan happened, the invocations carved from the scanner created tasks similar to the test case.  
The diagnostic is more actionable for us (clang module bug fixers), but I actually would categorize all `OutOfDate` diagnostics as that.  Related to the last tidbit in: https://github.com/llvm/llvm-project/pull/136612#discussion_r2054990077

https://github.com/llvm/llvm-project/pull/136612