Possible wrong order when selected ECMWF source

[Skealz]

In https://github.com/umr-lops/grdwindinversion/blob/main/grdwindinversion/inversion.py#L142

for ecmwf_name in ['ecmwf_0125_1h', 'ecmwf_0100_1h']:

Hence, the 'ecmwf_0125_1h' is tried first (and is selected if exists), but the source ecmwf_0100_1h seems to be more recent (and more precise) so it should be used. To be confirmed.

It is not that a big deal because it seems both doesn't cover the same period of time (seems to have a small overlap only).