wgsl: Add discussion about how OOB-ness is inherited

[zoddicus]

Covers the details of scalar, vector, and matrix operations.

Also includes a few minor cleanups to other sections of the doc that I made while reviewing the text.

Issue #3191

---

Requirements for PR author:

• All missing test coverage is tracked with "TODO" or .unimplemented().
• New helpers are /** documented */ and new helper files are found in helper_index.txt.
• Test behaves as expected in a WebGPU implementation. (If not passing, explain above.)

Requirements for reviewer sign-off:

• Tests are properly located in the test tree.
• Test descriptions allow a reader to "read only the test plans and evaluate coverage completeness", and accurately reflect the test code.
• Tests provide complete coverage (including validation control cases). Missing coverage MUST be covered by TODOs.
• Helpers and types promote readability and maintainability.

When landing this PR, be sure to make any necessary issue status updates.

[zoddicus]

I will review the CTS code and send up a seperate PR if there are code-side changes needed to conform with this discussion.

[jiangzhaoming]

Looks good to me, thanks!

I think the description added here is what we are already trying to emplace in the CTS implementation now, right? I think I can learn the general idea of simulating the possible computation steps (maybe multiple path) and flow the intermediate OOB to the result when following the implementation, and element-wise operation do such simulation in element-wise manner to get a possibly OOB-mixed result expectation.

[ben-clayton]

LGTM

[zoddicus]

Looks good to me, thanks!

I think the description added here is what we are already trying to emplace in the CTS implementation now, right? I think I can learn the general idea of simulating the possible computation steps (maybe multiple path) and flow the intermediate OOB to the result when following the implementation, and element-wise operation do such simulation in element-wise manner to get a possibly OOB-mixed result expectation.

The CTS is currently slightly inconsistent in this regard, but this is what was intended to be implemented. I have a WIP patch that fixes up a number of places that were incorrect that I should be sending up for review later today.