Fixup environment.yml

[dcamron]

A few changes I think will drastically speed up your environment generation.

1. Before Pythia "mystified", we had old theme dependencies that fell out of maintenance. These have conflicts with other major python and package versions and can drastically slow down environment generation. You might notice your environment has an older Python version than you expect.
2. PyTorch has retired support for their official pytorch conda channel. You can find community maintained pytorch-cpu and pytorch-gpu packages on conda-forge, or you can add the following to install from pip. Here I've opted for conda-forge.

# from environment.yml
# ...
- xbatcher
- pip:
  - torch

If you need to install from pip with more complex steps (specific wheels for specific drivers/hardware etc.) you can use pip install --no-cache-dir torch ... from within your conda environment.

[github-actions]

👋 Thanks for opening this PR! The Cookbook will be automatically built with GitHub Actions. To see the status of your deployment, click below.
🔍 Git commit SHA: 14cda22
✅ Deployment Preview URL: https://ProjectPythia.github.io/xbatcher-deep-learning/_preview/5

[s-kganz]

Ty for the guidance! We are up and running on development, but this will speed us along during the week.

[dcamron]

Let me know if this does help!

[s-kganz]

Can confirm this cuts the build time from ~5 minutes to ~2 minutes. Pulling the built image still takes some time, but a big improvement.