[Mitgcm-support] orlanski bug

[mitgcm-support at dev.mitgcm.org]

The orlanski bug is fixed. The bi,bj indices were out of bound because
some ENDDO's were in the wrong place in obcs_init_variables.F. Probably
my mistake. The model runs fine now on bonanza. 
Still uncertain (I tried some looping tests) why it worked just fine on
Sun's. 

In any case there is a new obcs_init_variables.F on bonanza in:
/u3/adcroft/SAMAR/mitgcm_latest/pkg/obcs/

There is also a new obcs_calc.F which fixes some superfluous ifdefs. 

Samar