trivial chance just to test push
alternative ice 'assimilation' scheme using bounds
T update from excess melting heat seems safe
only assign values in water points
better safe than sorry: avoid potential div by zero in melt_energy
fix: need ice before we can have snow on top of ice
wrong (and confusing) comment corrected
Merge branch 'winton_dev' of ssh://git.code.sf.net/p/getm/code into winton_dev