Improve clustering I

[trgardos]

A few suggested improvements to Clustering I lecture:

1. explain k-means++ since there seems to be time in this lecture
2. Remove text that says "k-means++" will be discussed in a future lecture. It is not AFAIK
3. Find better example for rescaling than horse kick data. Maybe build on age, income, gender type data.
4. Add another interesting clustering example at the end.