Escape control chars in source when printing diagnostics

[andyleiserson]

Replace control characters when emitting diagnostics in naga.

The motivation here is that my fuzzing runs were making my terminals unhappy. In most cases, in a browser, they won't go to a terminal so it won't matter, but this seems like the right thing to do regardless.

Testing
Adds a unit test and a wgsl_errors test.

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.

[ErichDonGubler]

LGTM minus nits!

[ErichDonGubler]

nitpick(non-blocking): I think it'd be easier to understand this code (including base += …) with base + found_pos factored out to a variable. Not a big deal, though.

[ErichDonGubler]

nitpick: We should use std::char::REPLACEMENT_CHARACTER instead of using this ad-hoc constant.

[andyleiserson]

I wanted to, but couldn't find a way to go from char to &str without a buffer.

[ErichDonGubler]

Oh, hmm. I suppose you could use a stack-allocated buffer and char::encode_utf8?

I'd be satisfied with merely a (doc) comment pointing to REPLACEMENT_CHARACTER expressing the type woe you're working around, too.

[andyleiserson]

I added a debug assertion that the character in the string equals char::REPLACEMENT_CHARACTER.

[ErichDonGubler]

suggestion: Let's add a CHANGELOG entry. This isn't likely to be noticeable for most folks' shaders, but it's good to call out.

[LukasKalbertodt]

I found this PR in the release notes and just wanted to drop this here: https://www.compart.com/en/unicode/block/U+2400
Unicode has special characters to display control characters. Might make sense to use those instead of FFFD for every control character.

[ErichDonGubler]

@LukasKalbertodt: Filed #8295. Thanks!