Add support for sweep gradients

[wmedrano]

Implement sweep gradient. Ported from Skia

• src/shaders/sweep_gradient.rs
  • SweepGradient::new - https://source.chromium.org/chromium/chromium/src/+/main:third_party/skia/src/shaders/gradients/SkSweepGradient.cpp;l=94;drc=4fac874565317b380eefeb3df759895a4c8a304c
  • SweepGradient::push_stages - https://source.chromium.org/chromium/chromium/src/+/main:third_party/skia/src/shaders/gradients/SkSweepGradient.cpp;l=88;drc=4fac874565317b380eefeb3df759895a4c8a304c
• src/pipeline/highp.rs: https://source.chromium.org/chromium/chromium/src/+/main:third_party/skia/src/opts/SkRasterPipeline_opts.h;l=3569;drc=d5c02fd369596fb4a6fa563f5506ff71ff7299ed
• src/pipeline/lowp.rs:
  • Not implemented. Could be done in a follow up PR if requested, but I don't have a need.
  • Tiny Skia will always use the highp implementation as xy_to_unit_angle is set to null_fn. If guard is here

[RazrFalcon]

Pretty clean. Looks good. Thanks! 👍

As for xy_to_unit_angle and LowP. It's fine to keep it HighP for now, but it seems like the code is nearly identical.
Up to you. We could add a comment like:

// TODO: can be implemented for lowp as well
null_fn, // XYToUnitAngle

[wmedrano]

Pretty clean. Looks good. Thanks! 👍

As for xy_to_unit_angle and LowP. It's fine to keep it HighP for now, but it seems like the code is nearly identical. Up to you. We could add a comment like:

// TODO: can be implemented for lowp as well
null_fn, // XYToUnitAngle

Ack, added comments. It requires a bit more code than it seems since we are missing some functions for f32x16 and lowp for concentric conical bias.

[RazrFalcon]

Got it. Makes sense.

[LaurenzV]

Haven't double-checked the implementation, but seems good to me.