Proper loops

[JMS55]

Connections
Closes #6521 and #4499

Description
Naga used to translate all for/while loops into while(true) loops, which hurt performance. This PR fixes that.

Disclaimer: Heavily LLM generated. I've done my best to review line-by-line, and I'm mostly confident in everything except the code that actually writes for/while loops for each backend.

Testing
Ran all existing naga tests, and eyeballed the generated shader diffs.

Squash or Rebase?
Squash

Checklist

• Run cargo fmt.
• Run taplo format.
• Run cargo clippy --tests. If applicable, add:
  • --target wasm32-unknown-unknown
• Run cargo xtask test to run tests.
• If this contains user-facing changes, add a CHANGELOG.md entry.

[JMS55]

I tested in a Bevy scene on Windows+Nvidia. Was not any faster.

Would love if more people could test on other platforms and shaders.

[mate-h]

GPU profiling data for mac reveals no conclusive results

Scenario	main	for-loops-test
1 frame many lights	15.95	15.60
1 frame many lights	16.08	15.58
1 frame many lights	16.17	15.83
1 frame many lights	15.89	15.84
1 frame many lights	15.84	15.85
5 frames many lights	79.43	78.63
5 frames many lights	79.00	79.00
5 frames many lights	79.17	82.40
5 frames many lights	79.33	81.66
5 frames many lights	79.15	79.07
SSR linear steps 128, bisection steps 32	48.92	48.98