Improve conversions between hydrolib and xarray dataset

[veenstrajelmer]

Follow up from: #527

Todo:

• assume FM quantity name in long_name attr in dfmt code, eg in plipointsDataset_to_ForcingModel and underlying functions. This makes uxuy exceptions less hard coded and accepts flexible variable names just like FM. We now use if hasattr(ds[datavar],'long_name') to rename the variables instead.
• consider locationname as var instead of attr to align singlepoint and multipoint netcdf dataset
• document _maybe_convert_fews_to_dfmt, restructure and explain why parts are necessary. Consider renaming. Not super important since it is a private function
• back-forth conversion also works for 2d timeseries, tidal bc and other cases? What happens in case of multiple quantities in bc file? Or quantities with unequal amount of points/nodes/times >> they will get nans since dims are aligned upon concatenation
• update example script/notebook to convert entire forcingmodel instead of only one forcingobj/point >> first support dataset of entire forcingmodel in Dataset_to_T3D?
• test ncbnd from workinprogress_interpolate_uds_toplipoints.py in fm run>> include unittest if none fails
• add nc writing as option to modelbuilder funtion and interpolate example script
• Add support for writing ForcingModel to netcdf instead of bc file HYDROLIB-core#318 (might not be necessary)
• fix sonarcloud code smells
• increase code coverage for new functions