Use non redundant form distance measures

[jolespin]

I checked out the tutorial and noticed you require the full redundant distance matrix which is way more memory intensive than it used to be. I want to try this out on my dataset with like 100k observations but a square form will cause my system to crash.