Add basic pyproject.toml for installing dependencies

[raehik]

Closes #9 . pip install -e . works as expected. CI failures are expected ( see #16 ).

Changes:

• remove some leftover IDE files
• enumerate dependencies in pyproject.toml
• add basic packaging template -- but not yet prepared as a library or package

[raehik]

(This is currently a draft and shouldn't be merged, but it looks like GitHub doesn't enable draft PRs in private repos.)

[raehik]

@jatkinson1000 installing deps with pip install -e . and running ./cmip26.py -85 85 -280 80 --CO2 0 --ntimes 10000 --factor 0 --chunk_size 50 --global_ 0, I'm getting the following error

ValueError: Bad Request: https://storage.googleapis.com/download/storage/v1/b/cmip6/o/GFDL_CM2_6%2Fcontrol%2Fsurface%2F.zmetadata?alt=media
User project specified in the request is invalid.

Is this down to missing an API key? I find it strange because I'm able to download that URL manually. (And it seems to be a GET, so either it's sending more info in the headers or my request differs to Python's in a bigger way.)

[jatkinson1000]

Yep, I can pass you Arthur's key which I have been using.
You also want to drop ntimes to 5 or 10 to stop it running on too much data.

[raehik]

This is tentatively ready for review. Worth noting:

• This only installs dependencies for running the cmip26.py script.
• Doesn't properly package -- mostly for installing dependencies.

[jatkinson1000]

In general looks good, and thanks for the work on this!
I have successfully pip installed and all built OK!
Didn't really test running, but that's hard to do at this stage.

Only real changes I'd suggest before merging would be to add some more metadata.
We might need a quick chat/check from @arthurBarthe to finalise these.

Left a few comments elsewhere but these are generally optional of for discussion.

[jatkinson1000]

We can probably write something tentative here, and/or as Arthur to provide.

[jatkinson1000]

Since we are imposing black we can add some of this into the file as follows:

[tool.black]
line-length = 88
target-version = ['py37']
include = '\.pyi?$'
exclude = '''
(
  /(
      \.eggs         # exclude a few common directories in the
    | \.git          # root of the project
    | \.hg
    | \.mypy_cache
    | \.tox
    | \.venv
    | _build
    | buck-out
    | build
    | dist
  )/
  | foo.py           # also separately exclude a file named foo.py in
                     # the root of the project
)
'''

This is optional though and I'm not sure I fully appreciate what it does at the moment.
Similar tools exist for linting etc.

[jatkinson1000]

It seems that stuff gets stored here after data is downloaded.
We will almost certainly change this when tackling #8 but is it worth setting to a specified local location for now?
Given the data seems to be stored after this script is run I'd also question whether tempfile.mkdtemp() is the correct command to use given the docs seem to imply it is for temp dirs that should be closed and removed before completion.
This probably isn't the place to deal with this, however.

[raehik]

As I understand, MLflow handles saving output:

mlflow.log_artifact(join(data_location, 'forcing'))

Which currently goes to ./mlruns/0/<hash>/artifacts/forcing. I think this is a tempdir and used correctly. Still useful to enable overriding (it was done so here), but I have a feeling tempfile can read envvars to do that.

[jatkinson1000]

Used correctly or incorrectly?
I didn't like the use of a tempfile to save important data when I saw it, and really tmpfiles should be cleared before ending the script.

I think you can probably just commit this as it should be dealt with in #4 and doesn't really matter where it creates it for now?