Time Agg Series Slot Config: Skip intervals containing NaNs [6036] [5]
Bug Number: n/a
Release notes (y/n): No (not this detail)
For Release Nums: 7.2

Merge of the Bug6036 branch (5 commits) to the RW 7.2 master branch.

A third aggregation option was added (presented as the second of three
choices) for the treatment of NaNs (or NaNs and Output values):

   "Do not aggregate intervals having NaN/Output values".

The target timesteps in the aggregated series slot of such intervals are
assigned NaN.
See draft document:
http://cadswes2.colorado.edu/~philw/2017/bugs/6036/SetTstepSize-SyncSlotsNanOptions-RW72-Enh.html

Current screenshots:
http://cadswes2.colorado.edu/~philw/2017/bugs/6036/2017-12-12/SetRunTimestepAction2.png
http://cadswes2.colorado.edu/~philw/2017/bugs/6036/2017-12-12/SynchronizeTimestepChangeDlg3.png
http://cadswes2.colorado.edu/~philw/2017/bugs/6036/2017-12-12/TimeAggSerSlotConfig3.png

Note: There is a pre-existing bug with the "Error on NaN" choice when the 
"First" or "Last"* aggregation function is selected.  Only one value in the 
source interval should be considered (i.e. the first or last), but we are 
generating an error if a NaN is encountered in some other timestep.  This 
preexisting problem isn't related to the introduction of the new "Do not 
aggregate intervals having NaN values" option, so I'm going to proceed with 
this code change, and file a bug for that problem.  DETAIL: The problem is 
that this condition is detected in the source timestep iteration in method 
SeriesSlot::TimeAggSpec::recomputeOwnerSlot(), but needs to be moved down 
to computeAggSlotValue() or the aggregation function helper functions 
called from that method.

*Note that "Last" is the implicit aggregation function for aggregation in 
the "Synchronize Timestep Change" operation when changing the model timestep 
size, e.g. from daily to monthly -- FOR VOLUME UNITS.