Add quantization library

[atvasilopoulos]

Description

Integrated a quantization library (based on this repo) in AIHWKit to extend its functionality for mixed-precision analog-digital hardware-aware training and simulations.

Details

New functionality includes:

• New tiles and arrays that support quantized periphery and quantized outputs
• An infrastructure to convert any model to a quantized counterpart with configurable activation/weight precision.
• Mixed-precision training with analog layers being trained in the conventional AIHWKit manner and quantized modules with conventional QAT methods
• Post-training quantization methods for any quantized module or tile, to avoid the necessity of retraining when possible

An example notebook is included that demonstrates the new functionality as well as docstring-based documentation of thew new methods and data structures.

NOTE: To adhere to black requirements, some edits were made to files independent of the current PR (merged by previous PRs)

[PabloCarmona]

Seems like the pipeline is stuck, can you update with master and try rerun the pipeline @atvasilopoulos ? Thanks!