Make duplicate handling better

[theianrobertson]

You can get at duplicate rows like:

comp.df1_unq_rows[comp.df1_unq_rows['acct_id'].isin(comp.intersect_rows['acct_id'])]

The compare report just says Any duplicates on match values: Yes. Could have some things like

• Count of duplicates (i.e. that weren't matched)
• More information in docs about how duplicates are picked (explain the algorithm)
• Shortcut on the class to get at the duplicates (like just shortcut the above?) Or maybe some way to point a discarded duplicate at the corresponding record it could have matched with?