Replicability test

[malerany]

Integrate functions to perform the replicability test from the scripts in project_management/scripts/replicability_tests/ifs-nemo_experiments/parallelized (located in the ifs-nemo-experiments branch of the project management repository) into the ReplicabilityTest class defined in src/pyhanami/diags/Replicability.py.

Assignees: @malerany

Migrated from GitLab: https://gitlab.earth.bsc.es/ces/hanami/pyhanami/-/issues/6

[malerany]

I implemented the methods within the ReplicabilityTest() class necessary for performing the replicability test. For now, the test is applied over the whole period available in the simulation datasets. However, it would be useful to allow the user to select a specific period.

Moreover, the computation of the scores is done in parallel using the concurrent.futures module. The number of workers used is set to None by default, but it can be controlled by modifying the MAX_WORKERS variables in the config.py file. The computation of the effect sizes and the application of the statistical tests are not parallelized for now.

The class is still missing the methods for checking that both provided ensembles are equivalent (in terms of variables, periods, etc.) and for creating a pdf report.

[malerany]

I created a new add_datasets method that allows adding new SimulationData objects after a ReplicabilityTest object has been initialised (it can be initialised with or without passing simulations to it). For example, we can do:

tester = ReplicabilityTest()
tester.add_datasets([ref,test])

or

tester = ReplicabilityTest(ref)
tester.add_datasets(test)

and all the simulations will be stored in tester.datasets. There is no limit on the number of datasets that can be added.

When calling the matrix plot function, the datasets are selected by passing the names of the SimulationData that should be included.

[malerany]

I also implemented a first version of the _compare_ensembles method that checks whether the provided ensembles have the same variables and lat-lon coordinates. This is done every time a new ensemble is added.