Need to reevaluate and refactor the way the configuration, scheduler and optimizer interact. For instance, I find confusing some checks on t_link_config when scheduling, because it disables the optimizer, but doesn't override with a number of actives.
In my opinion, scheduling should only be affected by the optimizer and the absolute storage limit (since one storage may be part of more than one link running).
All configuration related to number of actives and the like, should affect the optimizer. If the optimizer is disabled, leave the chosen value as is.
Other parameters may still be of use outside the optimizer (i.e. udt, ipv6...)