Update grid representations to use numpy instead of geopandas when considering a regular grid
An issue to track the updating of the grid representation to numpy.
In an attempt to increase speed. Currently the grid repr uses geopandas sjoin to join shapely polygons with points. This is good for any type of grid representation, but not the quickest method when considering regular grids.
As suh we can improve the aggregate function to only use numpy removing the need for pandas and geopandas https://earth.bsc.es/gitlab/ac/mapies/-/blob/master/mapies/grids/monarch.py#L131