Globbing of multiple file paths when reading model data

[jwarner8]

Currently in CSET the user provides a datapath, where dates can be swapped out for %Y%m%D, %N, etc., but there might be some issues when the user wants to pull specific data from an archive which contains different files. For example, at present the LFRic data is released as slammed (slam) and slammed and unified (slam_umf). As there are multiple output streams, using a wildcard * would include umified data.

For example, if someone just wanted to put the MASS paths, that could be an issue: a way round might be to accept multiple data paths for each model, like

%Y%m%dT%H%M_UK_1km_RAL3P2_lfric_slam_timeproc_%N_*.nc %Y%m%dT%H%M_UK_1km_RAL3P2_lfric_slam_diagnostics_%N_*.nc %Y%m%dT%H%M_UK_1km_RAL3P2_lfric_slam_pres_levels_%N_*.nc %Y%m%dT%H%M_UK_1km_RAL3P2_lfric_slam_model_levels_%N_*.nc
because
%Y%m%dT%H%M_UK_1km_RAL3P2_lfric_slam_*_%N_*.nc
would include anything with lfric_slam_umf_ in too and lead to lots of cube duplications.

A quick fix for this in RNS is if the data is umified, include umf in the filename, but not slam, as if it is umified it is most likely slammed. This will allow for easier globs in the short term, but in the future we might want to accept multiple paths.