filenames without timestamp

[paolap]

Currently we're counting on filenames to have some kind of timestamp that while different is at least consistent and allows to sort files in ascending time order.
This is broken when timestamp use jan, feb, mar etc for months.

This definitely cause issues in the following cases:

• mopdb when trying to work out frequency, and there's not frequency label in filename. It's fine if more than one timestep in file but it fails picking the right "next" successive file if there's only one timestep per file.
• mopper would fail selecting all the files using timestamp

It shouldn't cause issues when concatenating data as by default open_mfdataset() uses combine_by_coords

As the frequency in the above case can be adjusted manually in varlist, and if checking the timestamp fails there's the option of selecting files opening them and checking the time axis, both issue while slowing down the workflow should not cause errors. So for the moment I'm not going to add code to handle this case.