[Quantization] Allow loading of transform configs

[kylesayrs]

Purpose

• Support loading models with online transforms applied via Compressed Tensors (LLM Compressor)

Prerequisites

• [Transform] Support loading random hadamards on meta device vllm-project/compressed-tensors#445

Changes

• Require a minimum compressed tensors version of 0.11.0 (to support transform features)
• Load transform configs (if available), and apply them to the model before weight loading
• (misc) Refactor compressed tensors tests to check for perplexity, rather than exact output matches
• (misc) Remove update_dtype in order to reduce complexity and give users more control/predictability of model data types

Testing

• Regression tested using CompressedTensorsTest, added an online quip-style transformed model for testing
  • Perplexity results match expectations

[Rocketknight1]

cc @MekkCyber

[kylesayrs]

Putting in draft for now, need to do some more testing

[github-actions]

[For maintainers] Suggested jobs to run (before merge)

run-slow: compressed_tensors_integration

[kylesayrs]

#42887