Proposed features:

• refactor quantization codebase
• add true quantization (8bit) support for all methods
• implement AMP quantization (greedily for now) based on compression requirement

refer trailmet/algorithms/quantize/info.md for implementation details and remarks.
refer experiments/quantization/LAPQ/lapq_demo.ipynb for demo nb.

Do not merge until:

• update requirements to pytorch v2.0
• refactor quantization codebase to more simplified and modular
• add automatic mixed precision quantization feature
• add LAPQ true-quantization functionality
• test LAPQ true-quantization and add demo notebook
• test BRECQ fake-quantization with refactored changes
• add BRECQ true-quantization functionality
• test BRECQ true-quantization and add demo notebook
• update and test quantized inverted residual module for mobilenet
• fix logging
• check naming and linting
• update docstring and add comments