Add a module for implementing (sub-annual) time slices

[behnam-zakeri]

This PR introduces an algorithm for adding sub-annual time slices to a MESSAGEix scenario. This includes:

• Read the data from an input data (currently only Excel) file
• Configure the data of time slices for MESSAGEix sets (time slice names, duration, temporal level, temporal hierarchy).
• Identify technologies that must be represented at the time slice level.
• Add time slices to relevant parameters, e.g., "output", "input", "capacity_factor", "var_cost" etc., for relevant technologies.
• Solve the modified scenario.

How to review

There is a data file for the test case of Westeros baseline scenario. The reviewer can run the example code at the bottom of add_timeslice.py module. This can be done once for 2 and another time for 4 time slices, by changing the input parameter n_time. The results should show:

• ACT and demand for the respective time slices in each run.
• for 2 time slices, the results should be the same as "Westeros_seasonality.ipynb" tutorial.

PR checklist

• Continuous integration checks all ✅
• Add or expand tests; coverage checks both ✅
• Add, expand, or update documentation.
• Update doc/whatsnew.

[khaeru]

@behnam-zakeri —many thanks for this contribution!

As a first reaction, can you explain your thinking for contributing this code to message-ix-models rather than message_ix?

At a glance, the code seems to be sufficiently generic that it could apply to any MESSAGEix models, not only the ones in the MESSAGEix-GLOBIOM family. I say this because, for instance, the example in the if __name__ == "__main__": block applies the code to the Westeros model.

[behnam-zakeri]

@khaeru thanks for the point. Yes, indeed, this workflow can be applied to any MESSAGEix model. I agree this can be brought to message_ix as a tool.

[glatterf42]

Superseded by iiasa/message_ix#852, closing for now.