Speed up Image.filter() for I;16*

[akx]

Follows up on #8438; same split-by-mode optimization technique as in #9737.

Also fixes up things around ImageFilter:

• the tests' modes list was N-plicated
• the test_consistency_* tests didn't use the mode parametrization at all and just ran the same test N times
• test_crash claimed ImageFilter would crash on images smaller than the kernel; it doesn't, it just returns the same image.
  • Looks like test_crash had had assert_no_exception sometime before 3ec5059?
• The docstring for Kernel claimed only a couple of modes were supported.

[codspeed-hq]

Merging this PR will improve performance by ×2.4

⚠️ 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
✅ 343 untouched benchmarks
⏩ 2 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	test_filter[1024x1024-L-Smooth]	24.3 ms	10.1 ms	×2.4
⚡	test_filter[1024x1024-L-Sharpen]	24.2 ms	10.1 ms	×2.4

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 akx:faster-filter-still (525c545) with main (1cf685c)}

Footnotes

1. 2 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. ↩