PS obs not producing wind inc in EnVar

[rtodling]

It seems that in GSIBEC, for some yet unknown reason, when using pure ensemble (3D or 4D)-EnVar, we get wind increment from a surface pressure observation.

This needs investigation.

BTW: I don't recall whether GSI itself behaves in the same way - I've done this test before, but forget the result - need to do test again.

One thing I suspect is an update to the 3d-pressure fields (or its adjoint) that might not be taking place in the ensctl2state(ad) codes ...