Setup transient simulations to use "limit" rather than "cycle" or "extend" for historical stream files...

This was a problem for the CMIP5 simulations with CESM2 for the streams files: urbantv, pop-dens, and ndep for the future scenarios. With CESM2 the start date was 2015, but to be compatible with CMIP5 it needed to be 2006. So urbantv and pop-dens were were to extend so used 2015 values, while ndep cycled back to the future year of 2092. cycle and extend make sense for fixed 1850 or 2000 conditions, but for transient we should use "limit" so that it would abort if you got the start or end years wrong. "cycle" is especially bad in these cases.

This would also apply for lai-streams and prescribed soil moisture. Lightning right now is always a climatology so should always be cycle.