Improve trailing comma rule performance

[MichaReiser]

Summary

I'm not sure what precisely improves performance but these are the things I did:

• I removed a lot of the iterator chaining. Before flatten, filter_map, chain, coalesce, tuple_windows. Now it's just filter_map I would have expected that Rust sees through that but maybe it does not.
• I reordered the TokenType variants to match Tok.... I doubt that this has any impact
• I changed Token::from from &Spanned to (&Tok, TextRange).
• I avoid reading context.last() when we already set the context in the branch above

Test Plan

cargo test

[codspeed-hq]

CodSpeed Performance Report

Merging #9867 will improve performances by 10%

_{Comparing perf-trailing-comma (f7076c4) with main (daae28e)}

Summary

⚡ 10 improvements
✅ 20 untouched benchmarks

Benchmarks breakdown

	Benchmark	main	perf-trailing-comma	Change
⚡	linter/all-rules[large/dataset.py]	92.5 ms	84 ms	+10%
⚡	linter/all-rules[unicode/pypinyin.py]	10.8 ms	10.3 ms	+5.23%
⚡	linter/all-rules[pydantic/types.py]	43.9 ms	40 ms	+9.67%
⚡	linter/all-rules[numpy/ctypeslib.py]	20.6 ms	19.1 ms	+8.13%
⚡	linter/all-rules[numpy/globals.py]	2.8 ms	2.7 ms	+5.63%
⚡	linter/all-with-preview-rules[unicode/pypinyin.py]	11.8 ms	11.3 ms	+4.7%
⚡	linter/all-with-preview-rules[numpy/globals.py]	3.1 ms	2.9 ms	+5.02%
⚡	linter/all-with-preview-rules[numpy/ctypeslib.py]	23.3 ms	21.6 ms	+7.41%
⚡	linter/all-with-preview-rules[large/dataset.py]	105.1 ms	96 ms	+9.48%
⚡	linter/all-with-preview-rules[pydantic/types.py]	51 ms	47.4 ms	+7.77%

[charliermarsh]

I prefer your changes anyway lol.

[github-actions]

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

✅ ecosystem check detected no linter changes.