[clang] Correctly RAII reset two "new" Sema fields

[hahnjo]

FunctionScopesStart and InventedParameterInfosStart were introduced by upstream commit llvm/llvm-project@5c845c1 and are handled in ContextRAII, but not our ContextAndScopeRAII. The second commit reuses ContextRAII and should avoid this kind of problem in the future.

Could fix #14230, to be confirmed.

[phsft-bot]

Starting build on ROOT-performance-centos8-multicore/soversion, ROOT-ubuntu2204/nortcxxmod, ROOT-ubuntu2004/python3, mac12arm/cxx20, windows10/default
How to customize builds

[github-actions]

Test Results

    12 files      12 suites   2d 0h 27m 19s ⏱️
 2 605 tests  2 604 ✅ 0 💤 1 ❌
29 280 runs  29 279 ✅ 0 💤 1 ❌

For more details on these failures, see this check.

Results for commit fd66a49.

♻️ This comment has been updated with latest results.

[phsft-bot]

Starting build on ROOT-performance-centos8-multicore/soversion, ROOT-ubuntu2204/nortcxxmod, ROOT-ubuntu2004/python3, mac12arm/cxx20, windows10/default
How to customize builds

[hahnjo]

Independently from testing by ALICE, I believe this change is correct and I went ahead and synchronized it into a tag in our fork of the LLVM monorepo. @vgvassilev can you review please?

[vgvassilev]

This looks good to me. Do we have a plan how to move that upstream?

[phsft-bot]

Build failed on ROOT-ubuntu2004/python3.
See console output.

[hahnjo]

Do we have a plan how to move that upstream?

I was wondering the same. AFAICT we use it for autoloading, which will be hard to argue upstream. But by reusing ContextRAII, we can maybe get away with exposing a couple of public API functions and move it out of the Clang headers...