Phenology flags get locked in when starting. anew case using. prior case's restart

[ckoven]

I ran into a weird issue that I want to note here. TLDR is that the problem is fixed by #685, so once that is integrated it shouldn't happen any more, and if using older code and seeing similar behavior, merge in the #685 code.

Slightly longer description is that I was spinning up runs and then trying to take the restarts from the spinup run to start new simulations. But when I did that, I was seeing fundamentally different model behavior for the new cases than the spun-up dynamics of the prior case. It happened both when I set the restart file via a hybrid case or when I manually set the finidat argument in the user_nl_clm file. On discussing with @jenniferholm, @rgknox and others, we isolated the problem as one where the phenology flags were no longer updating in the new cases, and instead just remaining constant because of how the calendar logic interacted with the phenology logic. So plants that happened to have their leaves on at the time of the restart got infinite leaf lifespan and those that didn't got zero leaf lifespan. When I merged in the logic from #685, the behavior went back to what is expected.

[ckoven]

also noting that this is just a rediscovery of #627.