Issue with PartialCellBottom

[simone-silvestri]

I think I found a bug in the PartialCellBottom code.
The definition of volumes for ImmersedBoundaryGrids

Oceananigans.jl/src/ImmersedBoundaries/immersed_grid_metrics.jl

Line 26 in d289fed

@inline $volume(i, j, k, ibg::IBG) = $volume(i, j, k, ibg.underlying_grid)

bypasses the calculation of the volumes done in

Oceananigans.jl/src/Operators/spacings_and_areas_and_volumes.jl

Line 263 in d289fed

@inline $volume(i, j, k, grid) = $z_area(i, j, k, grid) * $z_spacing(i, j, k, grid)

Therefore, not accounting for reduced spacing in volumes. The same happens for vertical areas.
I will open a PR to fix it.

[simone-silvestri]

#3955 should have solved the issue with the partial cells.

The internal tide example looks like this on version 0.94.3:
https://clima.github.io/OceananigansDocumentation/v0.94.3/literated/internal_tide.mp4

In main, the results look much cleaner with less noise in the u velocity near the seamount:
https://clima.github.io/OceananigansDocumentation/dev/literated/internal_tide.mp4

However, I ll keep this issue open since we still need to add tests for the PartialCellBottom