Proposal: Drop Python 3.10 support

[aglinxinyuan]

The problem

Python 3.10 reaches end of life in October 2026 (devguide, already security-only). Our pyamber CI matrix still floors at 3.10 (["3.10","3.11","3.12","3.13"]), and that floor now blocks routine dependency updates.

Latest example — #6231: Dependabot bumped numpy to 2.5.1, which needs Python ≥ 3.12, so it wouldn't resolve on the 3.10/3.11 legs and CI went red. I had to cap it at numpy 2.2.6 (the last 3.10-compatible release) to unblock. This will keep recurring — numpy / scipy / scikit-learn all raise their Python floors on a schedule (NEP 29 / SPEC 0).

Proposal

Drop Python 3.10; make 3.11 the new floor (matrix → 3.11, 3.12, 3.13). 3.11 is supported until Oct 2027, so it's a safe floor.

Touches: .github/workflows/build.yml matrix, amber/pyproject.toml (Ruff target-version), bin/pylsp/Dockerfile, and the two setup docs.

This lifts our numpy ceiling from 2.2.x → 2.4.x. (Reaching numpy 2.5+ would also need dropping 3.11 — a separate, later step, since 3.11 isn't EOL yet.)

If there's no objection I'll open the PR.

[Yicong-Huang]

As 3.10 is reaching EOL in 3 months, and 1.3.0 won't be cut until Sept 1st, I agree we can remove its support. Please go ahead with the change on main (1.3.0-SNAPSHOT)! officially, v1.2.0-incubating will still support Python 3.10.