Adds bathymetry downscaling by averaging

[jagoosw]

This PR modifies regrid_bathymetry to take an argument method (instead of interpolaiton_passes) which is Interpolate(1) by default, and adds MedianAveraging which takes the average of the dataset depths within each (coarser) target cell.

This is with 20 interpolation passes:

The new method is almost smoother in the spiky bits and sharper on ridges.

I believe this is how they make the ORCA bathymetries: https://www.drakkar-ocean.eu/publications/reports/orca025-g70.pdf and this SCOR report claims median is better but still flawed for coarse models: https://scor-int.org/WG107Report.pdf

I haven't checked it works on GPU yet but hopefully.

[jagoosw]

This PR is a breaking change because it removes the argument interpolation_passes from regrid_bathymetry, should we support this and make the default:
...interpolation_passes = 1, method = Interpolation(interpolation_passes)) or just make a breaking change?

[codecov]

Codecov Report

❌ Patch coverage is 0% with 61 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/Bathymetry/regrid_bathymetry.jl	0.00%	61 Missing ⚠️

📢 Thoughts on this report? Let us know!