boundary condition changes

[kate]

set_tides fails to compile because mod_boundary has tests on NORTH_FSOBC for the definition of zeta_north, but set_tides doesn't have this test. I have a NORTHERN_WALL, so get:

 PGF90-S-0142-zeta_north is not a member of this RECORD (set_tides.f90: 60)
 PGF90-S-0142-ubar_north is not a member of this RECORD (set_tides.f90: 60)
 PGF90-S-0142-vbar_north is not a member of this RECORD (set_tides.f90: 60)

[arango]

Yes, we need to add the CPP conditionals here. It turns out that my test have four open boundaries so I didn't catch this bug.

[ocecept]

After the last revision (297) I have the same problem above. When using ana_fsobc and ana_m2obc with NOTHERN_WALL, get:
fortcom: Error: analytical.f90, line 136: This is not a field name that is defined in the encompassing structure. [ZETA_NORTH]

BOUNDARY(ng)%zeta_north(i)=0.0_r8

fortcom: Error: analytical.f90, line 433: This is not a field name that is defined in the encompassing structure. [UBAR_NORTH]

BOUNDARY(ng)%ubar_north(i)=0.0_r8

fortcom: Error: analytical.f90, line 436: This is not a field name that is defined in the encompassing structure. [VBAR_NORTH]

BOUNDARY(ng)%vbar_north(i)=0.0_r8

compilation aborted for analytical.f90 (code 1)

[arango]

Yes, thank you. I missed this one in ana_fsobc.h. I did took care of the others routines ana_m2obc.h, ana_m3obc.h, and ana_tobc.h when I made this update.

I also corrected missing RPM_RELAXATION conditional in w4dvar_ocean.h. Many thanks to Brian Powell for reporting this problem.