infra: Provide custom prefix to auto-related labels

[adswa]

In looking through the issues closed over the weekend, I felt compelled to act on the suggestion in #5833 to prefix a common, versioning-related identifier to the PR labels for the auto-workflow.
I think there was some general agreement in the original issue, but not enough momentum to implement the change. I personally keep forgetting which labels exist, and would be in favor of having them in one group to ease discoverability.
I'm suggesting semver as a prefix, because the corresponding test is called "Test for semver label". I'm happy to change this into anything else, though.

• if this change is accepted, the label names need to be changed

[codecov]

Codecov Report

Merging #6151 (229dd5f) into master (0df06be) will decrease coverage by 0.50%.
The diff coverage is n/a.

❗ Current head 229dd5f differs from pull request most recent head 349a021. Consider uploading reports for the commit 349a021 to get more accurate results

@@            Coverage Diff             @@
##           master    #6151      +/-   ##
==========================================
- Coverage   90.21%   89.70%   -0.51%     
==========================================
  Files         312      318       +6     
  Lines       42235    41854     -381     
==========================================
- Hits        38104    37547     -557     
- Misses       4131     4307     +176     

Impacted Files	Coverage Δ
datalad/_version.py	46.99% <0.00%> (-53.01%)	⬇️
datalad/support/due.py	48.00% <0.00%> (-32.00%)	⬇️
datalad/distribution/uninstall.py	71.42% <0.00%> (-28.58%)	⬇️
datalad/metadata/extractors/tests/test_exif.py	92.00% <0.00%> (-8.00%)	⬇️
datalad/metadata/extractors/tests/test_image.py	92.00% <0.00%> (-8.00%)	⬇️
datalad/interface/shell_completion.py	92.59% <0.00%> (-7.41%)	⬇️
datalad/distributed/create_sibling_gitlab.py	68.32% <0.00%> (-6.68%)	⬇️
datalad/metadata/extractors/tests/test_audio.py	93.54% <0.00%> (-6.46%)	⬇️
datalad/support/tests/test_due_utils.py	78.04% <0.00%> (-4.88%)	⬇️
datalad/interface/utils.py	91.36% <0.00%> (-3.96%)	⬇️
... and 85 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0df06be...349a021. Read the comment docs.

[yarikoptic]

I wonder on what disturbance that would cause interim since if we rename the labels and workflow in master -- maint from which we do cut releases wouldn't be able to auto release, so it would break that component.
If we are to do such a change, I feel it should happen right after some release so both maint and master could have a chance to auto-release without problems and collate perspective changelog based on labels of already merged PRs etc.

[mih]

I am all for such a change. I am totally unaware of what disturbances it would cause. But I do know that a release from master didn't work already with 0.15.0 (so no breakage possible here), and trying right after a release will only delay a reliable test of whether there is really no breakage. The worth case scenario -- an auto release does not work -- is nothing that concerns me at all.

[adswa]

With @mih's statement and @bpoldrack's thumbs up this kinda sounds like we could proceed here? AFAIK we could rename the current labels by editing them

[yarikoptic]

FWIW, on a 2nd thought, since auto analyzes PR labels only when running, and renaming (not creating new) hopefully would remain them in already labeled PRs, I think it should be safe to proceed but this PR most likely need to be done against maint, and then maint merged into master so everything is consistent and nice.

upon rename -- would be worthwhile to check recently merged PRs if their labels were updated correspondingly.

[mih]

Sound like a plan! Thx!

[adswa]

I rebased to maint, and will edit the labels now.

[adswa]

I had to re-attach the label to the PR to have it picked up by the semver test

[mih]

I killed the appveyor tests -- nothing in here is relevant for appveyor.

[mih]

OK, let's do it.