Configure cibuildwheel maximum Python version

[Jake Lishman (jakelishman)]

If we keep cibuildwheel completely up-to-date, it will attempt to build and test wheels up to the maximum current Python, including if it is in pre-release. We don't want our wheel-build jobs to fail if our dependencies aren't ready for the prerelease (as happened with dill during the Python 3.15 release-candidate period), so we want to skip the tests for them entirely. Our wheels are still compatible because we build for abi3.

I used CIBW_PROJECT_REQUIRES_PYTHON instead of messing with CIBW_BUILD or CIBW_TEST_SKIP because it's nicer to express the ABI intent with static globs on CIBW_BUILD (i.e. we build cp3??-* but not cp3??t-* as yet), and CIBW_TEST_SKIP is awkward because we'd have to calculate which versions of Python not to build; there's no "test-only" or the like.

Motivated by #16901

AI/LLM disclosure

• No part of this submission is LLM generated.
• Some written text was generated by:
• Some submitted code was generated by:

[qiskit-bot]

One or more of the following people are relevant to this code:

• @Qiskit/terra-core

[Coveralls (coveralls)]

Coverage Report for CI Build 34205988335

Coverage increased (+0.02%) to 87.873%

Details

• Coverage increased (+0.02%) from the base build.
• Patch coverage: No coverable lines changed in this PR.
• 4 coverage regressions across 2 files.

Uncovered Changes

No uncovered changes found.

Coverage Regressions

4 previously-covered lines in 2 files lost coverage.

File	Lines Losing Coverage	Coverage
crates/qasm2/src/lex.rs	3	92.8%
crates/circuit/src/parameter/parameter_expression.rs	1	90.31%

---

Coverage Stats

Relevant Lines:	132856
Covered Lines:	116745
Line Coverage:	87.87%
Coverage Strength:	985560.28 hits per line

---

💛 - Coveralls

[Jake Lishman (jakelishman)]

On hold while commit 9909985 is on the branch - that's just a test to make sure reconfiguring based on Python versions correctly restricts cibuildwheel.

[Jake Lishman (jakelishman)]

Excellent, it works completely as expected. See the old run, which was on top of 9909985: https://github.com/Qiskit/qiskit/actions/runs/34166990971?pr=16957.

[Matthew Treinish (mtreinish)]

Did it actually work correctly though? You specified the min as 3.11 and the max as 3.13 which should be >=3.11,<=3.13 but it only ran tests on 3.11 and 3.12. I would have expected it to test 3.13 too.

[Jake Lishman (jakelishman)]

Oh huh yeah, I misread the actual number of versions tested - I was only really paying attention to the lower bound... It'll be the <=3.13 specifier being interpreted as <=3.13.0, I bet. I could do the ugly hack of making it <=3.13.999 - that's probably actually easier than having to pipe it through a Python script or something just to add one to the minor version?