C++: Divide CODEOWNERS responsibilities.

[geoffw0]

Update CODEOWNERS responsibilities for the C team:

• previously all of cpp was assigned to @github/codeql-c-analysis.
• now directories are split between @github/codeql-c-analysis and @github/codeql-c-extractor (somewhat similarly to how the internal repo already is).
• note that later entries in the file override earlier ones.

As a member of the analysis but not extractor team I'm hoping to get slightly fewer notifications for things that aren't relevant to me after this change (e.g. changes to the autobuilder). The same would apply to @MathiasVP and @rdmarsh2. On the flip side, @AlexDenisov, @redsun82 and @sashabu may get slightly more (but relevant!) notifications as they are members of the extractor but not analysis team.

[jketema]

I don't think this is correct. downgrades and old-change-notes should be owned by analysis. That leaves the Windows autobuilder code, which indeed might be better assigned to extractor, although recently most work in there is actually done by the C# team.

[jketema]

Mmm, maybe that's not completely accurate either. At the very least I expect upgrades/downgrades/dbscheme to have the same owner.

[geoffw0]

downgrades and old-change-notes should be owned by analysis.

I feel like upgrades / downgrades are tightly coupled with extractor changes, so they should be owned by the extractor team. Arguably these kinds of changes are often interesting to everyone - so I could assign the entire c-team?

"old-change-notes" presumably is of interest to no-one. I don't know if I can/should express that. I certainly don't want notifying.

[jketema]

I feel like upgrades / downgrades are tightly coupled with extractor changes, so they should be owned by the extractor team. Arguably these kinds of changes are often interesting to everyone - so I could assign the entire c-team?

I'm not sure whether that's necessary. The "extractor" reviewer will generally be aware of this, because there's likely also a PR open with a relevant extractor change, which will refer to the CodeQL PR.

"old-change-notes" presumably is of interest to no-one. I don't know if I can/should express that. I certainly don't want notifying.

I think we still want to get notified when someone tries to change them by accident.

[geoffw0]

Adjustments made - everything belongs to analysis again apart from the autobuilder.

Sadly I can see this is only going to have a marginal effect on my notification spam (though an autobuilder PR was the motivating case for this change).

[jketema]

LGTM

[geoffw0]

Thanks. I'll allow a few hours for others to comment before merging...

[MathiasVP]

I totally agree with @jketema on this one. This PR is now a lot less controversial than it was initially, so this LGTM as well!