Add support for gridded datasets (and bilinear interpolation option)

[veenstrajelmer]

Is your feature request related to a problem? Please describe.
Currently, sample-based refinement works via a xyz list. For large datasets it can be much more computational efficient to supply the data as a grid and the lat/lon as a 1D array. Most bathymetry data is also stored in this format (like GEBCO with 7.5GB and EMODNET with 58x2GB). A dataset like that can also be used for bilinear interpolation.

Describe the solution you'd like
Add support for gridded datasets (and bilinear interpolation option)

Additional
Would be great if input could be a 1D array of lat, 1D array of lon and a 2D array of values. All floats per default, but it would be convenient if the latter also supports int datetype in order to save memory in for instance usage of GEBCO data.

[veenstrajelmer]

Fixed and merged in main. There is a notebook available with an example:
https://github.com/Deltares/MeshKernelPy/blob/main/docs/examples/05_mesh2d_refinement_gridded_samples.ipynb
use this for testing

[veenstrajelmer]

This is fixed with #29