Metal backend: fix bugs in qmv_impl

[manuelcandales]

Fixes two bugs in qmv_impl (also submitted PR 3096 to MLX)

• Fix 1: Changed qdot to qdot_safe when handling remaining elements in the N < 8 case
  Notice that in qmv_impl, there are two code paths based on whether N < 8:

  • When N >= 8: Correctly uses qdot_safe for remaining elements
  • When N < 8: Incorrectly used qdot instead of qdot_safe for remaining elements (fixed in this PR)

• Fix 2: Changed in_vec_size_g from floor division to ceiling division to correctly handle cases where K is not a multiple of group size

[manuelcandales]

Stack from ghstack (oldest at bottom):

• Metal backend: preserve meta data in decompose_linear_pass #17237
• Metal backend: fail test_modules.py if torchao can't be imported #17236
• Metal backend: Add Metal int4 quantization support to Parakeet #17235
• -> Metal backend: fix bugs in qmv_impl #17188
• Metal backend: import qmv_impl from MLX #17187
• Metal backend: int4 refactor checks and logs #17186
• Metal backend: test 4-bit linear with bias #17127
• Metal backend: enable quantization in test_modules #17118
• Metal backend: support 4-bit linear #17117
• Metal backend: remove parakeet linear decomp #17116
• Metal backend: enable linear with bias #17115
• Metal backend: test modules #17076

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/17188

• 📄 Preview Python docs built from this PR

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

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