Add support for two new env vars for chandra_models defaults + get_data info

[taldcroft]

Description

This adds two new environment variables that impact the behavior:

• CHANDRA_MODELS_REPO_DIR as an alias for THERMAL_MODELS_DIR_FOR_MATLAB_TOOLS_SW:
  override the default root for the chandra_models repository. The name defined by FOT Matlab tools is impossible
  to remember. 😄
• CHANDRA_MODELS_DEFAULT_VERSION: override the default repo version.

Adding, CHANDRA_MODELS_DEFAULT_VERSION environment variable might be controversial but this greatly facilitates two important use cases:

• Setting version to a fixed version in unit tests (e.g. with monkeypatch)
• Set to a developement branch to test a model file update with applications like yoshi where
  specifying a version would require a long chain of API updates.

This also adds a return dict of provenance information regarding the get_data call.

Interface impacts

Changes return of get_data from just the data to a tuple (data, provenance information).

Testing

Unit tests

• Mac

Independent check of unit tests by Jean

• Linux

Functional tests

@jeanconn extracted the FOT Matlab tools archive of chandra_models. This looks like below, where each directory is a shallow git repo with just the latest commit:

ska3-dev) ➜  chandra_aca git:(model-globs) ls /Users/aldcroft/Downloads/Thermal_Models
Archive/               chandra_models-3.43/   chandra_models-3.44/   chandra_models-3.46/   chandra_models-3.48/
chandra_models-3.41.1/ chandra_models-3.43.1/ chandra_models-3.45/   chandra_models-3.47/

Then I did the following in a dev ska with this repo installed:

In [11]: os.environ[
    ...:     "CHANDRA_MODELS_REPO_DIR"
    ...: ] = "/Users/aldcroft/Downloads/Thermal_Models/chandra_models-3.47"

In [12]: data, info = chandra_models.get_data("chandra_models/xija/aca/aca_spec.json")

In [13]: import json

In [14]: spec = json.loads(data)

In [15]: spec["name"]
Out[15]: 'aacccdpt'

In [16]: info
Out[16]: 
{'call_args': {'file_path': 'chandra_models/xija/aca/aca_spec.json',
  'version': None,
  'repo_path': 'None',
  'require_latest_version': False,
  'timeout': 5,
  'read_func': 'None',
  'read_func_kwargs': None},
 'version': '3.47',
 'commit': '5307bf8ad3a8588de3417b44f8eddf2a5a26a9e0',
 'data_file_path': '/Users/aldcroft/Downloads/Thermal_Models/chandra_models-3.47/chandra_models/xija/aca/aca_spec.json',
 'repo_path': '/Users/aldcroft/Downloads/Thermal_Models/chandra_models-3.47',
 'CHANDRA_MODELS_DEFAULT_VERSION': None,
 'CHANDRA_MODELS_REPO_DIR': '/Users/aldcroft/Downloads/Thermal_Models/chandra_models-3.47',
 'md5': '71cda7e5c2c5f72e5007f5d79014bd14'}

In [17]: del os.environ["CHANDRA_MODELS_REPO_DIR"]

In [18]: data, info = chandra_models.get_data("chandra_models/xija/aca/aca_spec.json")

In [19]: info
Out[19]: 
{'call_args': {'file_path': 'chandra_models/xija/aca/aca_spec.json',
  'version': None,
  'repo_path': 'None',
  'require_latest_version': False,
  'timeout': 5,
  'read_func': 'None',
  'read_func_kwargs': None},
 'version': '3.48',
 'commit': '68a58099a9b51bef52ef14fbd0f1971f950e6ba3',
 'data_file_path': '/Users/aldcroft/ska/data/chandra_models/chandra_models/xija/aca/aca_spec.json',
 'repo_path': '/Users/aldcroft/ska/data/chandra_models',
 'CHANDRA_MODELS_DEFAULT_VERSION': None,
 'CHANDRA_MODELS_REPO_DIR': None,
 'md5': '71cda7e5c2c5f72e5007f5d79014bd14'}

[jeanconn]

Feels a little odd to handle this via an "early" return but it is a short function.

[taldcroft]

From what I've read early returns are generally a good thing and encouraged to reduce code complexity. Here is an opinion from the internet that must therefore be true. http://www.itamarweiss.com/personal/2018/02/28/return-early-pattern.html

[taldcroft]

It's worth saying that this is something new-ish to me and might not be reflected in earlier code / patterns.

[jeanconn]

Changes make sense to me and unit and functional testing looks good.

[jeanconn]

I note that I think sot/ska_sun#31 needs to go at the same time.