[Mitgcm-support] Patrick: OBCS+CTRL inconsistencies.

[mitgcm-support at dev.mitgcm.org]

Before I forget, here's what I found today regarding OBCS as a part of
the control vector.

I can send these files when/if you want them.

ctrl_unpack: Now initializes ecco_ctrl for optimcycle=0.

ctrl_set_unpack_xz,yz : made it consistent with ctrl_set_pack, this
involved the nobcs and ncvarrecs loops once again.

ctrl_mask_set_xz,yz: the counting of wet points was inconsistent with
ctrl_init because of the overlap regions in x and y. Causes cbuffindex
to be incompatible and a runtime crash.

Right now, the pack and unpack routines seem to work. The optimization
routine will not work with obcs as a part of the control.

Jake