Improve benchmark test image

[akx]

The old image was highly channel-correlated, low-color, low-structure and non-representative of anything real. My bad in #9654 :)

The new image contains both low-energy, high-energy (full noise) and zero-energy segments.
If someone has an idea for an even better synthetic image, I'm all ears!

Let's see how the Codspeed numbers change - at least locally, test_quantize and test_histogram got a lot slower, because they actually need to do work 😄

Old	New

[codspeed-hq]

Merging this PR will degrade performance by 29.32%

⚠️ 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

⚡ 1 improved benchmark
❌ 1 regressed benchmark
✅ 343 untouched benchmarks
⏩ 2 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	test_quantize[1024x1024-RGB]	233.7 ms	527.8 ms	-55.72%
⚡	test_scale[1024x1024-RGB-0.01-NEAREST]	68.3 µs	60.5 µs	+12.82%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing akx:better-benchmark-image (69f73df) with main (72f6e83)}

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. ↩

[akx]

Merging this PR will degrade performance by 29.32%

As predicted! But no, code performance didn't change, just measurements for the more accurate...