Speed up Image.fill(), Image.linear_gradient() and Image.radial_gradient().

[akx]

Refs #9649, #9675, #9736 (same series, same techniques).

I didn't even know about linear_gradient and radial_gradient before looking at this...

[codspeed-hq]

Merging this PR will improve performance by ×2.1

⚠️ 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

⚡ 3 improved benchmarks
✅ 328 untouched benchmarks
🆕 10 new benchmarks

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	test_fill[1024x1024-RGBA]	4.6 ms	2.2 ms	×2.1
⚡	test_fill[1024x1024-LA]	4.6 ms	2.2 ms	×2.1
⚡	test_fill[1024x1024-RGB]	4.6 ms	2.2 ms	×2.1
🆕	test_linear_gradient[1]	N/A	76.4 µs	N/A
🆕	test_linear_gradient[F]	N/A	175.8 µs	N/A
🆕	test_linear_gradient[I]	N/A	175.7 µs	N/A
🆕	test_linear_gradient[L]	N/A	72.8 µs	N/A
🆕	test_linear_gradient[P]	N/A	95.4 µs	N/A
🆕	test_radial_gradient[1]	N/A	402.1 µs	N/A
🆕	test_radial_gradient[F]	N/A	524.3 µs	N/A
🆕	test_radial_gradient[I]	N/A	487.3 µs	N/A
🆕	test_radial_gradient[L]	N/A	398.2 µs	N/A
🆕	test_radial_gradient[P]	N/A	419 µs	N/A

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-fill (f7f350a) with main (6590b1b)}

[akx]

My local results for the new benchmarks:

linear_gradient was already well-optimized (the generated assembly did the right thing) for 1/L/P/F in the baseline, hence no gains over measurement noise there.

  -------------------------- benchmark 'allocate': 10 tests, 2 sources --------------------------
  Name (time in us)           0003_basegra OPS (Kops/s)  0004_newgrad OPS (Kops/s)  ΔOPS (Kops/s)
  -----------------------------------------------------------------------------------------------
  test_linear_gradient[1]                      575.2680                   581.8250          +1.1%
  test_linear_gradient[L]                      576.4618                   569.9319          -1.1%
  test_linear_gradient[P]                      421.5023                   400.9303          -4.9%
  test_linear_gradient[F]                      195.4028                   203.2991          +4.0%
  test_linear_gradient[I]                       39.9882                   201.5319        +404.0%
  test_radial_gradient[F]                       26.4618                    26.8236          +1.4%
  test_radial_gradient[L]                       22.2012                    26.1221         +17.7%
  test_radial_gradient[1]                       22.9234                    25.6476         +11.9%
  test_radial_gradient[P]                       22.1137                    25.4123         +14.9%
  test_radial_gradient[I]                       19.9459                    25.6850         +28.8%
  -----------------------------------------------------------------------------------------------