Add eigencalving option

[matthewhoffman]

This merge implements the “eigencalving” calving law described in: Levermann, A., T. Albrecht, R. Winkelmann, M. A. Martin, M. Haseloff,and I. Joughin (2012), Kinematic first-order calving law implies potential for abrupt ice-shelf retreat, Cryosph., 6(2), 273–286, doi:10.5194/tc-6-273-2012. In this law, calving rate is proportional to the product of the two principal strain rates if both are extensional.

The merge also makes a few required changes:

• reorganizes the calving module to make adding new options easier
• adds calculation of strain rates after velocity is calculated
• always call the RBF init routines (this had been only done in some circumstances as a workaround for an issue with a specific mesh)
• cells at floating margins that are thinner than their interior ice shelf neighbors are no longer consider dynamic, making a poor man's calving front advance scheme
• a bug is fixed in the extension of floating boundary conditions in Albany. Non-dynamic floating cells with ice were incorrectly being considered as the source for extension. They are now excluded.

Note the last two items change answers for runs with floating ice, but I confirmed the answer change is small, as expected. Runs with existing calving options will also have very slightly changed answers due to the reorganization of the calving module. These also appear to be very small.

[matthewhoffman]

@stephenprice , I propose we try to merge this soon to get the bug fixes it includes onto landice/develop. We could separate them into a different branch, but I think the eigencalving functionality is sufficiently working too - in my test of 100 years or so of AIS, it seemed to be giving sensible results. It would be good to have you do a quick review of the code if you have time. My hope is to merge this and most of the other open PRs later this week.

[matthewhoffman]

@stephenprice , I pushed a new commit that makes both corrections you identify. I'll merge the branch after I test it.