#17 Pairwise Distribution/Summary are inadequately tested

[Gaurav]

We need more testing for this code, as it has been
changing fairly erratically between versions. Testing
had been kept aside earlier as the code was extremely
old, mouldy, and in much need for repair. Since 0.9.1.
1, I've replaced PairwiseDistribution with nice, clean
code, which is a LOT less painful to deal with.
However, we need extensive testing to make sure that:
1. Numbers changing from last time are becoming
INCREASINGLY accurate, not less so.
2. They're not going to change *again* in the future
(except by becoming less accurate, which should be
protected by incorporating said tests into the Test
framework).

[Gaurav]

Logged In: YES
user_id=712450

Test 1: Ran a dataset (see attachment below). Replaced all
gaps with 'A's to prevent any overlap issues (I'll also run
another dataset from Guanyang tomorrow). I tested the
following:
1. Number of distances (based on the species composition
from SpeciesSummary): expected = obtained from export from
PairwiseSummary
2. Binning of distances: identical.
3. Overlap values (NOT size of overlap, but just the limits)
for both overlap and 5% cutoffs: identical.

I'll do test 2 (with Guanyang's dataset) tomorrow.