Resolved Issues in MARS/MIR

RESOL=AV uses wrong intermediate Gaussian grid

When using RESOL=AV and transforming from spectral to grid, the corresponding intermediate Gaussian grid to be used should be the cubic reduced Gaussian grid by default (i.e., O = T+1), as described above. However, at this point MIR uses a full (regular) gaussian grid with linear truncation (i.e., F = (T+1)/2).

Winds on rotated grids

There are known issues with wind fields when working with rotated grids. In particular, wind speeds are correct but the direction is wrong due to a problem in computing the angles.