Big optimizations for Buffer::set_rich_text

[UkoeHB]

Problem

Currently when Buffer::set_rich_text is called, all existing allocations in the buffer are simply discarded. This amounts to a lot of allocations for complex text.

Solution

Reuse existing allocations as much as possible when rebuilding a buffer.

Note that the changes here try to line up buffer reuse so it all maps back into the same positions in the buffer if possible. This way if, for example, you are just animating text color, the buffer won't actually require any extra allocations after the first time it's constructed (and won't over-allocate).

I did not originally intend for this PR to get so big, but diving into the code was a rabbit hole of allocations to squash.

Closes #301

Testing

I did not do any perf testing.

Follow-up

Additional areas for improvement:

• Buffer::set_text can also be optimized in a similar fashion.
• Use SmallVec instead of Vec where it makes sense. This should somewhat reduce the cost to make new buffers.
• There is one allocation at the top of set_rich_text that I didn't try to solve yet.
• Internally, ShapeLine is essentially a Vec<Vec<Vec<ShapeGlyph>>>. This can be flattened into two vecs: a vec of indices + metadata and a vec of ShapeGlyphs. Or even one vec with the metadata in-line.
• Internally, BufferLine caches a Vec<LayoutLine>, which is essentially Vec<Vec<LayoutGlyph>>. This could also be flattened.
• Buffer caches a ShapeBuffer but it would be better to have this in FontSystem or elsewhere for global reuse.

[jackpot51]

The Rust/build CI will need to pass

[UkoeHB]

@jackpot51 fixed. Looks like cargo-deny needs to be updated on your end.