Update out-of-the-box initial conditions files by running them through init_interp

[billsacks]

With the changes in #311 , out-of-the-box initial conditions files are incompatible with the memory allocation, so use_init_interp always needs to be true. This should be okay for now, but we can get a speedup in initialization by pointing to already-interp'ed initial conditions files. (This will clearly give a speedup for cases where we no longer need to use init_interp, but I think it will also give a speedup when we need init_interp, because new restart files are about 1/3 the size of old ones.)

In talking about this with @ekluzek , we decided that I probably shouldn't do this interpolation as part of #311 . The main reason is that I want #311 to be bit-for-bit. When we do the interpolation, we figure we might as well move to a gx1v7 mask (most initial conditions files still have a gx1v6 mask). I think that making this change in mask could change answers for cases that do an additional interpolation from this file (imagine a lake point near the Caspian Sea: the source point for the interpolation could differ now).

So we feel that the interpolation of initial conditions should happen in a separate tag.

[billsacks]

Note when doing this: the finidat pointed to by the interp_f19_noncrop testmod should also be updated (changing it to interp_f09_noncrop if necessary).

[billsacks]

Maybe this can be addressed along with #313

[billsacks]

We may or may not do this for the cesm2 release, depending on time.

[ekluzek]

I'm pretty sure we said we'd do this for cmip6 now.

[billsacks]

We'll still do this on master, but not on the release branch