Add new generic setup methods for VectorModel subclass

[hboisgon]

New generic methods to implement (names to be discussed, each method will be split into separate issues):

• [x] setup_vector: prepare your response_units object from region (from shapefile or from subbasin with uparea/pfaf threshold). This object should have has a property that links polygons within a specific order.
• [ ] setup_forcing_from_"constant" (uniform timeseries without georeferencing): date selection + time interpolation
• [ ] setup_forcing_from_geodataset: date selection + time interpolation + geographic interpolation (nearest/point-to-point/mean at first, maybe a max buffer to which need to fill with something else or interpolate the few stations to grid) + optional zonal statistics (if need grid in between)
• [ ] setup_forcing_from_raster: date selection + time interpolation + zonal_statistics
• [ ] setup_vector_data_from_raster: zonal_stats*
• [ ] setup_vector_data_from_mapping: reclass + zonal stats
• [ ] setup_constant_pars: one static value for all units
• [ ]

Other functionalities:

• [ ] interpolate missing data on geodataset
• [ ] expend geodataset to polygon/line
• [ ] define proper geodataset object for self.response_units
• [ ] *zonal_statistics_percent as well as classic mean/sum/mode...
• [ ] USDA soil texture from percent clay,silt and sand (equation is currently in wflow --> hydromt_wflow/workflows/ptf.py)
• [ ] E_{ref} from grass equation to add

[DirkEilander]

Above is a long list with wishes. I think we should start with the following methods:

• [ ] setup_vector_data_from_raster. this method should read a rasterdataset; perform user defined zonal statistics (e.g., fraction, mean, max, min, mode, etc) based on self.vector; save using set_vector
• [ ] idem for forcing.

Note separate issues on zonal stats in #167 and #597