Bump torchao pin to pick up qmv_fast optimization

[manuelcandales]

Performance improvements to lowbit quantized linear metal kernels in torchao. See AO PR#2167 for details.
The table below summarizes torchchat's decode speed (tokens/second) on Metal backend on M1 Max 64GB after this update

# bits	Llama 3.2-1B	Llama 3.2-3B	Llama 3.1-8B
1	179.96	87.01	51.10
2	186.91	98.13	54.37
3	170.62	85.69	48.05
4	175.15	89.54	50.51
5	147.10	70.19	38.58
6	140.51	63.62	35.48
7	131.27	64.19	32.69

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/torchchat/1541

• 📄 Preview Python docs built from this PR

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

❌ 1 New Failure, 3 Unrelated Failures

As of commit 0698a1d with merge base a37b08a ():

NEW FAILURE - The following job has failed:

• pull / test-build-runner-et-android / linux-job (gh)
  RuntimeError: Command docker exec -t 8f05e606be25d5248ca329f8bfacbfd7e9855aeb20a62dccfa0b5478722d0c63 /exec failed with exit code 1

BROKEN TRUNK - The following jobs failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / runner-et (16-core-ubuntu) (gh) (trunk failure)
  ##[error]The operation was canceled.
• pull / runner-et (macos-14-xlarge) (gh) (trunk failure)
  Process completed with exit code 1.
• pull / test-torchao-experimental-et (macos-14-xlarge) (gh) (trunk failure)
  Process completed with exit code 1.

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

[Jack-Khuu]

You can ignore the ET issues I'm bumping it today

lmk when you want to land and I'll bypass/force