vello_cpu: Support custom atlas size

[fundon]

Currently, the atlas size is not customizable and DEFAULT_GLYPH_ATLAS_SIZE can cause excessive memory usage.

And expose Resources::before_render Resources::after_render methods, so this part of code can be simplified.

The behavior of vello_cpu Resources and vello_hybrid Resources will be more consistent.

[LaurenzV]

I can get behind configuring the atlas size, but not why you want to expose before_render/after_render? This is supposed to be abstracted away from the user.

[LaurenzV]

Oh, I see now this is in parley… I believe this is still using an older version of vello CPU, so the code hasn’t been updated yet. It should be possible to remove all of this.

[fundon]

Another scenario is that after rendering, we can pass atlas.get(glyph_key), and then we can upload the atlas to our custom GPU texture.

[fundon]

It should be possible to remove all of this.

I hope there is a more concise way. After rendering, cache it into GlyphAtlas, and subsequent users can easily obtain data from GlyphAtlas::get.

[LaurenzV]

LGTM overall.

[LaurenzV]

By the way, I do want to warn you that the atlas caching API is still experimental and not meant for production use yet. While it seems to work, there are still a number of issues to resolve, especially around performance when zooming.

[fundon]

By the way, I do want to warn you that the atlas caching API is still experimental and not meant for production use yet. While it seems to work, there are still a number of issues to resolve, especially around performance when zooming.

Ok. I'll give feedback if there are any problems. Thanks.