modelbuilder fix inexact latlon bbox

[veenstrajelmer]

After merging #1061, the modelbuilder downloads CMEMS data for a too small region. This is due to the fact that meshkernel does not stick to the exact bbox when generating the mesh. bbox code from dfmt.generate_bndpli_cutland():

import dfm_tools as dfmt
import numpy as np

crs = 'EPSG:4326' # coordinate reference system
lon_min, lon_max, lat_min, lat_max = 105.8, 106.85, 17.75, 18.5
dxy = 0.05

# generate spherical regular grid
mk = dfmt.make_basegrid(lon_min, lon_max, lat_min, lat_max, dx=dxy, dy=dxy, crs=crs)

# get bbox
mesh_bnds = mk.mesh2d_get_mesh_boundaries_as_polygons()
if mesh_bnds.geometry_separator in mesh_bnds.x_coordinates:
    raise Exception('use dfmt.generate_bndpli_cutland() on an uncut grid')
mesh_bnds_xy = np.c_[mesh_bnds.x_coordinates,mesh_bnds.y_coordinates]
bbox = (mesh_bnds.x_coordinates.min(), mesh_bnds.y_coordinates.min(), mesh_bnds.x_coordinates.max(), mesh_bnds.y_coordinates.max())
print(bbox)

Prints (xmin, ymin, xmax, ymax):

(105.8, 17.75, 106.85, 18.529811688024356)

So the max latitude is larger than requested.

Todo:

• move bbox code to separate function
• fix meshkernel imports in code so all classes are imported directly >> no, since we also use mk.__version__
• overwrite lat/lons in modelbuilder notebook with meshkernel bbox
• also fix in modelbuilder example script >> no, discontinue in discontinue modelbuilder example python script #1068