Good luck with that document I'm sure it made more sense when I posted it :-)
One other thing on the Data Page LIO. In general that is calculated in general for non-unique nonclustered index access as :
(data row cluster ratio * filter rows) / rows_per_page_for_table) + ((1 - data row cluster ratio)*filter rows))
If the estimate page count derived from the rows tends toward 1 then I think the estimate datapage lio count is simply the number of rows (as Jeff/Kevin pointed out).