add missing mul! methods with transpose

[ExpandingMan]

This adds the missing mul! methods that break the initial (minimal) example in #55. The * method now falls back to this instead of having a separate definition with mutating NNlib.scatter.

It is undoubtedly inefficient to make a separate call to fill! before executing the NNlib kernel, but I have verified that NNlib already does this, so this shouldn't be any different than using scatter.

PR Checklist

• Tests are added
• Documentation, if applicable

[mcabbott]

Thanks!

I think some tests on GPU are in order, as the ones here pass before the PR. (Which is still good, to check that it doesn't break things which worked.)

Also tests for various ways to provide the wrong-sized arrays to this method. (IIRC NNlib.scatter! might be @inbounds.)

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 95.18%. Comparing base (0ea288d) to head (f22da3d).
Report is 5 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #56      +/-   ##
==========================================
- Coverage   96.26%   95.18%   -1.09%     
==========================================
  Files           3        3              
  Lines         134      166      +32     
==========================================
+ Hits          129      158      +29     
- Misses          5        8       +3     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[mcabbott]

Let's do it.