Minor upgrades to bit pack

[vayuda]

Now allows the packing dimension to be -1
No longer need to declare the device to use in the function params as it takes it from the input data
Added rudimentary benchmark comparing with fp16 and using unpack in a weight only quantization model

• faster for input vectors of size 4096 to 8192 but slower for sizes outside of that range (would like a review on the setup for another pr though)

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/347

• 📄 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

As of commit bfd6c5f with merge base 950a893 ():

NEW FAILURE - The following job has failed:

• Run Regression Tests / test (CUDA Nightly, linux.g5.12xlarge.nvidia.gpu, --pre torch --index-url https://download.pytorc... / linux-job (gh)
  RuntimeError: Command docker exec -t b0479fb5d008243ce47d20fb16ba5e4e5cdbf94c6d93c622669167cd785af471 /exec failed with exit code 1

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