Devel/refactor

[oloapinivad]

This is a simple reorganization of the ecmean library and the relative functions. it is cleaner although now all the different functions are scattered around.

[oloapinivad]

This closes #51

[oloapinivad]

Last commits introduce a new way of parsing, closes #61

[oloapinivad]

Last commit has made a revolution, creating an ecmean package to be installed. this means that now we have to

1. create the conda env with mamba env create -f environment.yml
2. activate the environment conda activate ECmean4
3. install the package pip install -e .

With the current version, we can even run global_mean from command line!

[oloapinivad]

This closes #63, #64, #66

[oloapinivad]

This closes partially #65

[oloapinivad]

I found a small error in the way the cmor mask were handled (% instead of fraction) which is now solved. I also tried to introduce the oceanic mask. Now testing the computational performance.

[oloapinivad]

Test on levante suggest that there still a bit of confusion on two aspects

1. The mask unit: we cannot use the one that is shipped within the file, often absent or badly formulated. The solution of dividing by 100 the cmor is the only one that look safe
2. The recognition of regular/curvilinear/unstructured grid is everything but failsafe. However, since we plan to roll to CDO based interpolation, I will wait for this

One reflection regards the components. Each of them require a lot of manual handling. I am wondering if it will be a good idea to introduce a defaultatm and defaultoce component in line with CMOR, and then use exceptions from this

[oloapinivad]

@jhardenberg please let me know if you have time to test this. I would like to close and proceed this week.

[oloapinivad]

Introducing coverage (around 90%) to be published on coveralls.io and some more elaborated test for command line arguments. Merging soon.