Fix ImageChops.offset for 16-bit modes + speed up

[akx]

Follows up on #9753 (via @hugovk's excellent catch #9753 (review)).

This replaces the whole internal implementation of ImageChops.offset with a much simpler memcpy() pattern rather than computing modulos for every pixel...

[codspeed-hq]

Merging this PR will improve performance by ×2.8

⚠️ 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

⚡ 4 improved benchmarks
✅ 599 untouched benchmarks
⏩ 335 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	test_offset[1237x811-L]	11 ms	1.9 ms	×5.7
⚡	test_offset[1237x811-LA]	16.7 ms	7.4 ms	×2.3
⚡	test_offset[1237x811-RGB]	16.7 ms	7.4 ms	×2.3
⚡	test_offset[1237x811-RGBA]	16.7 ms	7.4 ms	×2.3

Tip

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

---

_{Comparing akx:offset-16-fix (ed627f5) with main (3c7080c)}

Footnotes

1. 335 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. ↩

[radarhere]

I've created akx#28 with some suggestions.

[akx]

Oh, I realized there's a much better way to do this still, roundaboutly via @radarhere's suggestion PR...

[akx]

-Merging this PR will improve performance by 16.37%
+Merging this PR will improve performance by ×2.8

There we go... 😄