APCEMM interface

[thabbott]

I'm leaving this PR as a draft for now because I'm anticipating that I'll make changes as I get this interface ready for the pycontrails workshop. Some likely changes include

• Exposing additional YAML parameters in the low-level interface (pycontrails.models.apcemm.interface) to give more control to "expert" users
• Adding options to the high-level interface as needed to support case studies for the pycontrails workshop
• Documenting known limitations, if any come up in conversations with @sdeastham

But, I think the interface is at a point where it's ready for review. @mlshapiro: go to town. @nickmasson: would you mind looking specifically at the run function in interface.py and the run_apcemm method in adapter.py to make sure I'm not doing anything dumb setting up child processes to run APCEMM?

@sdeastham: feel free to chime in here as much or a little as you'd like!

Changes

Features

• Add interfaces for running APCEMM from within pycontrails. The interface includes:
  • a low-level wrapper around the native APCEMM interface (pycontrails.models.apcemm.interface) than allows users to generate APCEMM input files and run a compiled APCEMM binary manually.
  • a high-level adapter (pycontrails.models.apcemm.APCEMM) that allows users to run APCEMM through the pycontrails Model interface.
• Add APCEMM tutorial notebook.

Internals

• Create unit tests for the APCEMM interface that run if an APCEMM executable is found on the PATH inside a clean APCEMM git repository with a pinned git hash. These tests will be skipped unless users carefully configure their local test environment and will not run in CI.
• Exclude APCEMM tutorial notebook from notebook tests.

Tests

• [x] QC test passes locally (make test)
• [x] CI tests pass

Reviewer

@mlshapiro @nickmasson

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[thabbott]

I've finished making changes--@mlshapiro, you're free to test drive again!

I will note that the output in the notebook looks different than what you saw initially. I think this is because I found and fixed an issue in the azimuth initialization in DryAdvection, and the contrail is being spread less by normal shear as a result.