Drop Python 3.11 and raise dependency minimums

[nabobalis]

PR Description

This drops Python 3.11 and raises the minimum dependency versions for 2.5, in line with sunpy:

• Python >= 3.12. This updates tox, CI, ruff, the RTD environment and cruft. oldestdeps now runs on 3.12 and the Windows job on 3.13. SPEC 0 drops 3.12 in October, so we might revisit this at release time.
• astropy >= 7.0.0
• matplotlib >= 3.10.0
• dask >= 2024.6.0 and specutils >= 1.16.0 (tests-optional)
• numpy stays at >= 1.26 and scipy doesn't change.

Cleanups:

• Removes the "numpy 1" repr in test_resampled_wcs.py. The difference really comes from astropy < 6.1.3. With numpy 1.26 and astropy >= 7, the version gate picks the wrong string and oldestdeps fails.
• Removes the astropy < 7 branch in test_compound_wcs.py.
• Removes a stale numpy 2 TODO in utils/cube.py. The code was already changed to np.prod.
• Deletes the unused pre-cruft .sunpy-template.yml.
• Points the gilesbot figure report URLs at the py314 jobs that CircleCI actually runs.

Closes #731

AI Assistance Disclosure

AI tools were used for:

• Code generation (e.g., when writing an implementation or fixing a bug)
• Test/benchmark generation
• Documentation (including examples)
• Research and understanding
• No AI tools were used

Regardless of AI use, the human contributor remains fully responsible for correctness, design choices, licensing compatibility, and long-term maintainability.

K

[Cadair]

Thanks. I think I'll sit on this until we are closer to 2.5 to make backports a bit easier.

We are keeping numpy at 1.26 (over all the sunpy packages) until it causes us trouble just for people who might still be stuck.

[nabobalis]

I squashed this into one commit on top of main. Since the last version, numpy stays at 1.26 as you said. astropy is 7.0.0 instead of 7.0.2, and scipy and the tests-only/docs extras are back to what main has. I also dropped the global pyparsing filter in pytest.ini and kept the existing oldestdeps pin. Keeping numpy 1.26 with astropy 7 made the numpy 1 repr gate in test_resampled_wcs.py pick the wrong string, so I removed that too.