First-draft infrastructure for evaluation

[thanasibakis]

This PR's goal is to take a first stab at how our package will provide model evaluation functionality.

Key changes

eval subpackage

The package now has an eval subpackage. The idea here is that this will contain modules for the various classes of metrics we will want (e.g. on proportions directly, on samples, on lineage domination times). I have a simple MAE on proportions in the eval.proportions submodule to start us out.

demo

The exploration/demo workspace has two model-fitting scripts now. The idea here is that the outputs of each are collected into some output folder, and pitted against each other. exploration/demo/evaluate.py does this, but hardcodes CSV paths for now.

Other changes

• The utils submodule continues to grow with helper functions that I think are nice to have all in one place. Happy to hear other ideas here.
• I will slowly move away from the R dependency for visualizing things. The visualize submodule has a new function for plotting posterior proportion trajectories over time, replacing the R script I had before.
• I pushed a very small change to the simulation study script to resolve a todo, which could have waited for another PR but snuck into this one accidentally.
• I also somehow managed earlier to commit some scratch CRPS code to main at some point, so that's why you'll see linmod/eval.py deleted here.

[afmagee42]

I'm encountering a few issues that are outside the scope of this PR, but which are blocking me from checking that the code runs in my newly-cloned repo. I suggest that this PR is merged as soon as @swo approves, and I will make a new branch to see about cleaning those up.

[swo]

I'm blocked right away:

linmod-py3.10➜  demo git:(first-metric) ../../data/load_metadata.py > metadata.csv
Will load from https://data.nextstrain.org/files/ncov/open/metadata.tsv.zst

Downloading...Traceback (most recent call last):
  File "/home/ulp7/proj/cfa-viral-lineage-model/exploration/demo/../../data/load_metadata.py", line 108, in <module>
    data = load_metadata(
  File "/home/ulp7/proj/cfa-viral-lineage-model/exploration/demo/../../data/load_metadata.py", line 58, in load_metadata
    with urlopen(url) as response, save_path.open("wb") as out_file:
  File "/usr/lib/python3.10/pathlib.py", line 1119, in open
    return self._accessor.open(self, mode, buffering, encoding, errors,
FileNotFoundError: [Errno 2] No such file or directory: '.cache/data.nextstrain.org/files/ncov/open/metadata.tsv'

[afmagee42]

@swo see #12