Fix to test job mem against MaxMemPer[CPU|Node] limits at scheduling time.

Initially job mem limits were tested at submission time through
_validate_min_mem_partition() -> _valid_pn_min_mem(), but not tested
again at scheduling time, thus leading to jobs incorrectly being scheduled
against partitions where the job exceeded their MaxMemPer* limit
(which can in turn be inherited from the system wide limit too).

NOTE: New WAIT_PN_MEM_LIMIT job_state_reason enum component added to support
this new waiting reason.

Bug 2291.