WIP: Standardized runner func

[delgadom]

• closes #xxxx
• tests added / passed
• docs reflect changes
• passes flake8 climate_toolbox tests docs
• entry in HISTORY.rst

An attempt at creating a single standardized runner function. Intended to be paired with climate_transform_specs PR !5

Proposal

Include the full load, transformation, aggregation, and writing phases as argument to a main aggregation function. Different loaders and transformations should be able to handle all variants of transformations, including loading multiple variables, multiple outputs, etc.

In this proposal, a call would look like the following:

transform_and_weighted_aggregate_climate_data(
    climate_data_loader=load_and_standardize_climate_dataset_from_pattern,
    loader_kwargs={
            'pattern': READ_PATH,
            'source': 'GMFDv1',
            'read_variable': 'tas'},
    transform=tas_poly,
    extra_transform_kwargs=None,
    aggregator=weighted_aggregate_grid_to_regions,
    extra_aggregation_kwargs={
        'variable': 'tas',
        'agglev': 'hierid',
        'aggwt': 'popwt'},
    validate=validate_tas_poly}],
    extra_validation_kwargs={
        'valid_tas_range': (-20, 55),
        'assert_no_nans': True},
    writers=[
        atomic_netcdf_write_to_pattern,
        fgh_header_write_to_pattern]}],
    extra_writer_kwargs: [
        {'pattern': WRITE_FILE, 'ext': 'nc4'},
        {'pattern': WRITE_FILE, 'ext': 'fgh'}],
    assign_attrs_list=[
        'source',
        'variable',
        'agglev',
        'aggwt',
        'valid_tas_range',
        'year'],
    year=1980,
    variable='tas-poly-3',
    power=power,
)

In a parameterized format, this could look like the following:

from toolz import curry as prepackage_keyword_arguments

tas_poly_gmfd = prepackage_keyword_arguments(
    climate_toolbox.transform_and_weighted_aggregate_climate_data,
    climate_data_loader=load_and_standardize_climate_dataset_from_pattern,
    loader_kwargs={
            'pattern': READ_PATH,
            'source': 'GMFDv1',
            'read_variable': 'tas'},
    transform=tas_poly,
    extra_transform_kwargs=None,
    aggregator=weighted_aggregate_grid_to_regions,
    extra_aggregation_kwargs={
        'variable': 'tas',
        'agglev': 'hierid',
        'aggwt': 'popwt'},
    validate=validate_tas_poly}],
    extra_validation_kwargs={
        'valid_tas_range': (-20, 55),
        'assert_no_nans': True},
    writers=[
        atomic_netcdf_write_to_pattern,
        fgh_header_write_to_pattern]}],
    extra_writer_kwargs: [
        {'pattern': WRITE_FILE, 'ext': 'nc4'},
        {'pattern': WRITE_FILE, 'ext': 'fgh'}],
    assign_attrs_list=[
        'source',
        'variable',
        'agglev',
        'aggwt',
        'valid_tas_range',
        'year'])

PARAMETERIZED_KWARGS = [
    [{'year': y} for y in common.GMFD_v1.MODEL_YEARS],
    [{
            'variable': ('tas-poly-{}'.format(power) if power > 1 else 'tas'),
            'power': power}
        for power in range(1, 6)],
]

if __name__ == '__main__':
    slurm_runner(job_spec=(PARAMETERIZED_KWARGS))(tas_poly_gmfd)