[Impeller] Snap glyph positions to screen space pixels and map UVs correctly

[bdero]

Resolves flutter/flutter#121887.

Hoo boy, this one took longer than I'd like to admit to figure out. Improvements:

1. Properly snap the vertex positions to the screen space pixel grid by keeping track of screen space.
2. Glyph positions are converted to screen space and rounded; glyph sizes are converted to "local" (no translation) screen space and rounded up.
3. UVs are properly mapped to the atlas with a half texel offset so that we sample in the middle of texels.
4. Only use these rounding behaviors when the transform is translation-scale (which is when we nearest sample). In particular, 2 only works for affine matrices.

I think we can come through and remove more CPU cycles here, but wanted to land this since it finally fixes the problem where we skip sampling rows/columns in glyphs.

Before:

Untitled2.mov

After:

Untitled1.mov

[flutter-dashboard]

It looks like this pull request may not have tests. Please make sure to add tests before merging. If you need an exemption to this rule, contact Hixie on the #hackers channel in Chat (don't just cc him here, he won't see it! He's on Discord!).

If you are not sure if you need tests, consider this rule of thumb: the purpose of a test is to make sure someone doesn't accidentally revert the fix. Ask yourself, is there anything in your PR that you feel it is important we not accidentally revert back to how it was before your fix?

Reviewers: Read the Tree Hygiene page and make sure this patch meets those guidelines before LGTMing.

[jonahwilliams]

LGTM!

[flutter-dashboard]

Golden file changes have been found for this pull request. Click here to view and triage (e.g. because this is an intentional change).

If you are still iterating on this change and are not ready to resolve the images on the Flutter Gold dashboard, consider marking this PR as a draft pull request above. You will still be able to view image results on the dashboard, commenting will be silenced, and the check will not try to resolve itself until marked ready for review.

Changes reported for pull request #40912 at sha 4784512

[chinmaygarde]

This golden issue seems related

[bdero]

Once the text gets far enough away (I believe when it passes behind the singularity) it seems to flip to the other side and start coming inward.

We used to apply this same transform to the position on the GPU, but now we're doing it on the CPU, so I suspect something is overflowing or not getting clamped correctly in our matrix utils on the CPU. Looking into it.

[bdero]

Figured it out. I was doing a vec2 transform instead of a vec4 transform, and so we were resetting the projective coord prior to the final transform to clip space.