can we get rid of rotation on import of imod5 models?

[luitjansl]

Currently when we import imod5 models we always get an ANGLE defined in the npf package, even when the major and minor axes of the conductivity ellipsoid are aligned with the grid axes. But when that's the case, we can swap the names of K and K22 and then we don't need the angle anymore. This simplifies calculations on the modflow side.

This issue comes from a review comment by @Huite:

Just a note, maybe you're aware already: iMOD5 chooses the north (or 90 degrees with x-axis) as the principal direction of the horizontal conductivity. MODFLOW 6 chooses the x-axis (0 degrees) as the principal direction (as you'd expect from the unit circle). If I recall correctly, in the iMOD5 conversion, this results in there always being an 90 degree angle defined even if there's no anisotropy. However, this makes the formulation step of MODFLOW 6 more expensive, because it requires trigonometry evaluations. It's worth considering, I think, to consistently transform the iMOD5 conductivity to an x-axis aligned one for k11.

[JoerivanEngelen]

Will be followed upon in #1397