The isochrone option just pushes an isochrone out. Obviously, the time steps are more regular than the grid option. It is slower, but there are many reasons it should be better - I only use the isochronal option now.

In each weather model, the parameters in each column are effectively an average at that level. WRF uses a quadrilateral cell, ICON triangular, some are hexagonal etc. The wind components are at the boundaries.

The 10m winds are derived quantities from the wind components and the atmospheric structure.

The grib data you get is often interpolated to a regular grid. Exceptions where it is on the original grid are HRRR, RAP, Exp WRF etc.

Exp then needs to interpolate the data. It fits a smooth surface in space and time to the data, which should generally be better than a linear interpolation. It adds computational over-head, but I think is worth it.