Feat/add docker support

[anson70242]

Summary

This PR adds Docker support to provide a standardized and highly reproducible execution environment for the D3R project. This is crucial for ensuring that the specific environmental dependencies (PyTorch 1.11.0 / CUDA 11.3 / Python 3.9) can be consistently and easily deployed across different machines.

Key Changes

1. New Dockerfile:
   • Uses pytorch/pytorch:1.11.0-cuda11.3-cudnn8-runtime as the base image.
   • Crucial Optimization: Core dependencies (like numpy, pandas, scikit-learn, etc.) are installed directly via pip install within the Dockerfile to prevent common installation conflicts associated with locking versions in requirements.txt for this specific Python 3.9 environment.
2. Updated README.md:
   • Introduces a concise "If you prefer Docker..." block, providing clean instructions for building and running experiments, including the necessary volume mount (-v .:/app/data) for easy dataset access.

[anson70242]

It's tested to be working.