Remaining tests for the new precision recall

[michaelbornholdt]

Here is a list of ideas on what tests to add to the precision recall function. This is needed because #63 is only focussed on results and not on good tests.

• Use a small example as Florian has written in precision_recall() counts within group connections twice but not between group connections #62 to make sure that the math is correct
• Make sure that the output df has the right shape depending on the number of groupby_cols
• Add checks that make sure the groupby is available, ie it exists in the df.

Test which is about replicate groups, not about groupby_cols

• Make sure that if the replicate groups span a unique mapping over the input df, then all values should be 0. This is the case because no connection will have a true label.