fix slow performance of dfmt.open_dataset_curvilinear()` due to incorrectly added time dimension

[veenstrajelmer]

This code has terrible performance since a time dimension is added to vertices_longitude` and other vars:

import dfm_tools as dfmt

file_nc = r'P:\archivedprojects\11206304-futuremares-rawdata-preps\data\CMIP6_BC\CMCC-ESM2\vo_Omon_CMCC-ESM2_historical_r1i1p1f1_gn_*.nc'
uds = dfmt.open_dataset_curvilinear(file_nc, convert_360to180=True)#, data_vars="minimal")

When passing data_vars='minimal' this is avoided. Add this directly to the xr.open_mfdataset() call in the function