Update code

[munhouiani]

• Create conda environment files for mac and linux
• Create option to disable joblib when n=1 while running dataset creation
• Use Pcap iterator to avoid out of memory error
• Remove petastorm and use huggingface load_dataset
• Reduce number of epochs to 20 and use all cores for Dataloader
• Fix seed for reproducibility
• Remove dockerfile