Computes the duration of the available data in the request intervals that are both volatile and partial for the
given table.
If we computed the duration of the intervals in the table that are just volatile and present, then the
comparator will erroneously favor those tables that have a larger volatility range.
Suppose we have two tables t1, and t2. t1 is monthly, and t2 is daily. t1's volatile intervals are
[01-2015/01-2016] while t2's volatile intervals are [12-2015/01-2016]. t1 is missing the bucket
[12-2015/01-2016], while t2 is missing the bucket[12-31-2015/01-01-2016]. Suppose we make a request for
[01-2015/01-2016] at the monthly grain. Both tables have equal volatility and partiality at the request grain,
and t2 has more data available in the partial-but-volatile range. However, a comparator that looked just at
volatility would favor t1, because it has 11 months of volatile-but-present data, while t2 only has 30 days of
volatile-but-present data, even though t2 actually has more present data.