James Liu

As a final sanity check, I reran microbenchmarks for this PR. Gains are generally pretty small or within error margin, with the exception of `Query::get` and `World::get` for sparse components....

Does this count as a breaking change? The bindings for the shader do seem like a public interface that should be governed by semver.

Conservatively assigning this to 0.11 for now, feel free to move to 0.10.1 if this shouldn't be considered breaking.

Definitely seems like there's a significant gain here. Microbenchmark results below: ``` group main reduced-branching-bundles ----- ---- ------------------------- add_remove/sparse_set 1.01 751.5±37.73µs ? ?/sec 1.00 746.1±41.38µs ? ?/sec add_remove/table 1.00 1122.5±18.47µs...

> It's interesting that all of the benchmarks are a bit faster, except for the spawn_world ones which are all consistently a bit slower. I'm not sure where this is...

Even if the texture formats were supported, compute shaders are not supported on the GL backend, which the `game_of_life` example is built on. I don't think there's any way around...

Pinging @JonahPlusPlus about this. Not sure how much of bevy_atmosphere we can upstream, but having it be first-party, well integrated with rest of the first-party renderer, and supporting the base...

> * If we split up the main pass node, we could add the skybox as a separate node between the opaque node and the next, in a separate plugin,...

@JMS55 can you update the PR description? This is difficult to understand what this PR is doing and it's action plan from just the PR title.

I think the problem with `Rect` is the need to return glam types (i.e. Vec2, IVec2, UVec2), as well as the associated type level assertions needed for each one potentially...