Refactor test suite

[padix-key]

Resolves #694 .

This PR refactors tests/ and benchmarks/. The changes are mainly aesthetic (with random seeds being the exception) the test logic has not changed:

• Replaced itertools.product in parametrizations with multiple stacked @pytest.mark.parametrize decorators
• Added ids for parametrizations with opaque or long-path arguments
• Switched from os.path to pathlib.Path (including glob/splitext); data_dir() now returns a Path
• Moved PDB-derived structure files into tests/structure/data/pdb
• Replaced load_structure()/save_structure() calls with the respective File classes
• Replaced manual tempfile handling with the tmp_path fixture
• Normalized from pytest import approx to pytest.approx
• Replaced legacy global np.random.seed with local np.random.default_rng
• Converted comma-separated parametrize argname strings ("a, b") to lists (["a", "b"])

[codspeed-hq]

Merging this PR will not alter performance

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 2 improved benchmarks
❌ 1 (👁 1) regressed benchmark
✅ 90 untouched benchmarks
🆕 10 new benchmarks
⏩ 14 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	benchmark_match_kmer_selection[KmerTable-None]	278.5 µs	243.3 µs	+14.49%
⚡	benchmark_match[KmerTable-None]	326.9 µs	294.9 µs	+10.87%
🆕	benchmark_get_sequences[cas9.fasta]	N/A	8.5 ms	N/A
🆕	benchmark_get_sequences[ec_bl21.fasta]	N/A	168.1 ms	N/A
🆕	benchmark_lddt[all-False]	N/A	206.3 ms	N/A
🆕	benchmark_lddt[all-True]	N/A	869.6 ms	N/A
🆕	benchmark_lddt[atom-False]	N/A	290.4 ms	N/A
🆕	benchmark_lddt[atom-True]	N/A	1.9 s	N/A
🆕	benchmark_lddt[chain-False]	N/A	296.5 ms	N/A
🆕	benchmark_lddt[chain-True]	N/A	1.9 s	N/A
🆕	benchmark_lddt[residue-False]	N/A	301.8 ms	N/A
🆕	benchmark_lddt[residue-True]	N/A	1.9 s	N/A
👁	benchmark_match_kmer_selection[KmerTable-11*11*1*1***111]	245.1 µs	276.1 µs	-11.22%

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing padix-key:issue-694 (e7abbe3) with main (00054e5)}

Footnotes

1. 14 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[padix-key]

The branch of this PR is branched off #921, since this PR also introduced some test changes in test_pdb.py, test_pdbx.py and test_mol.py. So these changes can be ignored in this PR and the PR will be rebased on main once #921 is merged.