Initialize more arrays below kmt

[mnlevy1981]

Running call_compute_subroutines with the intel compiler using the following
FFLAGS: -g -O0 -init=snan,arrays crashed until I initialized several fields
below the sea floor.

Leaving this is draft status until @mvertens verifies that it fixes the error she was seeing.

Fixes #371

[mnlevy1981]

It would be useful to figure out what gfortran flags can be set in order to detect this error, then our continuous integration can ensure similar problems don't slip back in.

[mnlevy1981]

Also, there might be similar problems with the ciso code but we don't have a test for that

[mnlevy1981]

Verified this is sufficient to get @mvertens's test to pass

[mnlevy1981]

@klindsay28 as it turns out, the -finit-real=snan flag isn't sufficient to trap unitialized real variables in a derived type. I added -finit-derived per some gfortran documentation, except it includes the note

These options do not initialize

• objects with the POINTER attribute
• allocatable arrays
• variables that appear in an EQUIVALENCE statement.

And these are allocatable arrays. So when cheyenne is back up I'll add -init=snan,arrays to the intel flags, verify it's aborting when I don't initialize hflux_out and sflux_out, and then use it to see if I also need to initialize remin (currently marbl_interior_tendency_share_set_used_particle_terms_to_zero() does not set sinking_particle%remin(k) = c0 because I was too trusting of the gfortran flags, but I suspect I'll need to add that line back in). Then I'll test in CESM and, assuming all tests pass, merge this to development.

[mnlevy1981]

Per the previous comment, I tried running the stand-alone test with just hflux_out and sflux_out set to 0 below the sea floor, and the intel compiler didn't like that. So dd7cb41 does two things:

1. sets remin to 0 below the sea floor
2. updates the compile flags for running the stand-alone driver with intel

I tested that commit on cheyenne by running aux_pop_MARBL with both intel and gnu -- both sets of tests were bit-for-bit identical with baselines, so I'll merge this in and make a new tag.