Make wgpu_render_pass_set_bind_group clamp the number of dynamic offsets within limits.

[bradwerth]

This similarly updates wgpu_compute_pass_set_bind_group. Neither function throws an error when clamping. It's not clear what mechanism there is for error reporting at this entry point.

Connections
N/A

Description
User agents calling wgpu_compute_pass_set_bind_group directly can provide an arbitrary number of dynamic offsets without passing validation.

Testing
Untested. Any test using something like cpass.set_bind_group is stopped by validation.

Checklist

• Run cargo fmt.
• Run cargo clippy. If applicable, add:
  • --target wasm32-unknown-unknown
  • --target wasm32-unknown-emscripten
• Run cargo xtask test to run tests.
• Add change to CHANGELOG.md. See simple instructions inside file.

[cwfitzgerald]

For validation we should include the original length, so that in the body of the pass we can check to make sure the original_length == clamped length, and error if they don't match.

[teoxoy]

Briefly looking at the code around num_dynamic_offsets it looks to me like it would make more sense to just change its data type to be usize since every time we set it we cast it to u8 and every time we read it we cast it back to usize.

[cwfitzgerald]

For bookkeeping, going to request changes

[nical]

Addressing this differently in #5026