delete old torchao to_nvfp4 triton kernel

[vkuzo]

Summary:

we moved to mslk in #4031,
deleting the old kernel

Separate PR to keep PRs small

Test Plan:

pytest test/prototype/mx_formats -s

[vkuzo]

Stack from ghstack (oldest at bottom):

• -> delete old torchao to_nvfp4 triton kernel #4078

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/4078

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 Cancelled Job, 4 Pending

As of commit cf3e267 with merge base 5c0c814 ():

CANCELLED JOB - The following job was cancelled. Please retry:

• Run Regression Tests (gh)

This comment was automatically generated by Dr. CI and updates every 15 minutes.