[Backport maintenance/2.15.x] Fix deprecated-method false positive

[github-actions]

Backport 57f38c3 from #7795.

[Pierre-Sassoulas]

First back-port that worked, the benefit is that we don't have to do the cherry-picking manually and we will always be able to release with minimal work because the maintenance branch will be ready. But we do have to add the labels and approve the PR.

[Pierre-Sassoulas]

@DanielNoord, I excluded the primer on the maintenance branch, what's your opinion about that ?

[DanielNoord]

@DanielNoord, I excluded the primer on the maintenance branch, what's your opinion about that ?

I think we should probably still run it. The cherry picking might introduce hard to spot bugs because of dependent commits we don't know about.
Seeing as it wouldn't run that often it also doesn't cost us too much CI time.

[Pierre-Sassoulas]

Testing the configuration added to main / maintenance/2.15.x with #7815

[Pierre-Sassoulas]

@DanielNoord it seems the primer can't run on the maintenance branch right now:

 Archive:  primer_commitstring.zip
  inflating: commit_string_3.11.txt  
  inflating: commit_string_3.7.txt   
cp: cannot stat 'commit_string.txt': No such file or directory

What do you think ?

[coveralls]

Pull Request Test Coverage Report for Build 3530420845

• 1 of 1 (100.0%) changed or added relevant line in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-0.0005%) to 95.34%

---

Totals
Change from base Build 3527537560:	-0.0005%
Covered Lines:	17083
Relevant Lines:	17918

---

💛 - Coveralls

[jacobtylerwalls]

I wouldn't bother running the primer on the maintenance branches. If we've merged work to main, it's far too late to be parsing and investigating if changes are false positives or intended changes on the support branches.

[Pierre-Sassoulas]

Also if the primer caugth something it's going to be added to our functional tests before being merged so there's less risks.

[DanielNoord]

I suggested it initially as it might catch merge conflicts that are not directly exposed by the tests. However, if it is such a hassle then let's not.

[Pierre-Sassoulas]

I wanted to disable some more checks but it requires a refactor because on: needs to be defined for the whole file and I want to keep the pylint check in "Check". This will wait for 2.16's maintenance.