#17 Persistent - Mean update

[nlang]

At end-of-replication a DSOL persistent's value gets
updated one last time. The current update value used
is the persistent's MEAN. However, this ought to be
the persistent's LAST VALUE.
In addition, I propose to adapt 'getSampleMean
()/getSampleVariance()/etc...' so they update the
values (by using a dummy update with 'last value')
before returning. Internal updates should decrease N
to meet expected behavior (i.e. N is the number of
recorded updates, thus excludes internal updates).

[nlang]

Logged In: YES
user_id=747091

The previous report WRONGLY assumed the persistent (i.e.
nl.tudelft.simulation.dsol.statistics.Persistent) gets
updated at end-of-replication. On closer inspection, this
turns out to be incorrect: a summary tally gets updated at
end-of-replication, NOT the persistent itself. Using the
sampleMean to update the summary tally is in itself
correct.
However, the issue remains that the recorded sampleMean is
not corrected for time elapsed since the last event. The
last part of the bug-report should thus still be addressed.

[nlang]

Logged In: YES
user_id=747091

The fix suggested in the initial report (dummy update) may
seriously affect variance and thus std. dev measurements. A
quick alternative would be to provide access to the time of
the last recorded event. In this way, users have at least
the option to correct the final stats in a way they deem
reasonable.

[nlang]

Logged In: YES
user_id=747091

Using TWO dummy events (rather than one) also fixes
variance statistics. N and Sum properties should be
adjusted to mask dummy updates.