Fixing the issue that negative sublimation was not considered

Fixing the issue (#772) that negative sublimation was not considered when not full elements were sublimating off, when the module vapour transport was not enabled.

It turns out it also strongly affects the disappearance of surface hoar. See below. The upper one is from the current master branch, the bottom one from this branch. This needs to be tested before merging. It looks like the master branch overpredicts surface hoar formation.