AWQ QuantizationMixin + SequentialPipeline

[brian-dellabetta]

SUMMARY:

• Add QuantizationMixin to AWQModifier so we don't have redundant inputs (num_bits, symmetric, group_size)
• Move AWQModifier to sequential pipelines, to avoid huge memory requirements of caching all activations at once.

Regression test results are acceptable, results are all roughly the same, and within stderr, see test plan below.

Resolves #1409
Resolves #1369
Related to #1383
Related to #1406
Related to #1368
Related to #1410

More improvements split into #1435

TEST PLAN:

• Rerun tests to validate
  No regression in tests, comparing against those reported in original AWQ PR. All gsm8k results are within stderr:

Type	gsm8k	wikitext
Old AWQ+QuantModifier Sym	.1054, .1069	9.1931
New AWQ+QuantMixin Sym	.1077, .1084	9.1841
Old AWQ+QuantModifier Asym	.1274, .1281	9.0281
New AWQ+QuantMixin Asym	.1312, .1350	9.0288

[github-actions]

👋 Hi! Thank you for contributing to llm-compressor. Please add the ready label when the PR is ready for review.

Note: This is required to complete the testing suite, please only add the label once the PR is code complete and local testing has been performed.

[kylesayrs]

This modifier still has extreme memory constraints due to how we're still megabatching everything for _set_module_kwargs and _apply_smoothing

This followup PR gets part way towards removing that, but still batches outputs in its current state (grep for torch.cat): #1435

Let's land this as is and add accumulation in a follow up to reduce memory requirements and weird kwarg assumptions

[rahul-tuli]

Lgtm! Good updates!