Set ndte=240 and other pre release updates

[apcraig]

update CICE version,
set ndte=240,
fix uninitialized forcing field box2001,
update icepack

• Developer(s): tcraig

• Are the code changes bit for bit, different at roundoff level, or more substantial? roundoff.

Test results at https://github.com/CICE-Consortium/Test-Results/wiki/cice_by_hash_forks, hash 0f6f079 shows all tests except boxadv debug pass on conrad with 4 compilers (as expected). 63/67 tests are NOT bit-for-bit as expected (alt01 and boxdyn are bfb). No QC testing done due to confidence in implementation of ndte change.

• Does this PR create or have dependencies on Icepack or any other models? Uses updated icepack hash but icepack is bit-for-bit.

• Is the documentation being updated with this PR? (Y/N) Y
  If not, does the documentation need to be updated separately at a later time? (Y/N) N

• Other Relevant Details:

[apcraig]

Once this is merged, I will run the full "weekend" test suite manually on all platforms to establish new baselines and verify it's working fine on all machines.

[eclare108213]

The initialization of cldf looks a little out of place, but that array is unique to this module so it's handled differently from the others. Ok.

[apcraig]

Thanks for the quick reviews and merge. I have started launched test suites on multiple platforms and will follow up tomorrow.

Regarding the cldf initialization. I agree 100% it's awkward. It's a quick fix to work around a boxdyn debug failure with the cray compiler. I thought about a few other implementations, but decided this was the most direct and least risky. I think with regard to the forcing options, we need to think about what variables should be set by the forcing and if there are some that are not set by some of the forcing options, what should happen. We should probably require that all forcing fields be set to something for each forcing option, even if there is no data. It's also partly why I ran a bunch of tests with the signaling nans today on conrad. I wanted to get an idea of how bad we were with uninitialized variables, see #247. That's probably something we should have done a while ago, but didn't. It's not as bad as I feared, and I think can be deferred until after the release. It looks like the default options are fine and that bgc and box, maybe not surprisingly, need a little work.

[eclare108213]

When we replace the LYq forcing, we might be able to get rid of cldf completely.