Is your feature request related to a problem? Please describe.
Distance measures are currently not tested, both python or cython implementations. This leads to bugs like #645 and #596

Describe the solution you'd like
For every distance measure, a corresponding unit test should be made to check known inputs against known outputs. This will a) test whether the distance measures are working properly and b) run checks in travis ensuring any new development does not silently break / alter the distance measure code / behaviour.

Describe alternatives you've considered
No alternatives afaik.

Additional context
This will help with any classifier which uses distance measures, as if the distance measure pass all tests but your classifier does not you know the problem lies in your classifier rather than the distance measures. Likewise, the reverse is true if a problem lies in a distance measure instead of your classifier.