curvilinear_compute_rectangular_grid_on_extension does not result in grid with constant dx/dy

[veenstrajelmer]

Describe the bug
When generating a spherical mesh I would expect the edge length to be relatively constant. I request 4/60 degrees, which is approximately 40075/360/60*4=7.42 km. However, the edge lengths of the resulting grid vary from 3340 to 5515. This might be valid, but also the edge length varies over latitude which I think should not be the case.

To Reproduce

from meshkernel import MakeGridParameters, MeshKernel, ProjectionType
import dfm_tools as dfmt

#general settings
lon_min,lon_max,lat_min,lat_max = -8, 3, 46, 52
lon_res,lat_res = 4/60, 4/60 # 4 nautical miles, is approx 40075/360/60*4=7.42 km

figsize = (10,8)
crs = 'EPSG:4326'

# Create an instance of MakeGridParameters and set the values
make_grid_parameters = MakeGridParameters(angle=0.0,
                                          origin_x=lon_min,
                                          origin_y=lat_min,
                                          upper_right_x=lon_max,
                                          upper_right_y=lat_max,
                                          block_size_x=lon_res,
                                          block_size_y=lat_res)

mk = MeshKernel(projection=ProjectionType.SPHERICAL)
mk.curvilinear_compute_rectangular_grid_on_extension(make_grid_parameters)
mk.curvilinear_convert_to_mesh2d() #convert to ugrid/mesh2d

#write xugrid grid to netcdf
xu_grid_uds = dfmt.meshkernel_to_UgridDataset(mk, crs=crs)
netfile = 'test_grid_no_constant_edge_length.nc'
xu_grid_uds.ugrid.to_netcdf(netfile)
#TODO: grid has edge_length of 3340 to 5155 meters, should this not be more constant?

Expected behavior
A resolution of about 7400 meters and more constant over space. First also check if the actual DCSM grid has varying edge length over latitude.

Screenshots

Version info (please complete the following information):

• OS: Windows 10
• Version 3.0.0