Implement Rhythm Patterns, Rhythm Histogram and Statistical Spectrum Descriptor

[dbogdanov]

According to some feedback from their creators, the combination of RP + SSD + RH is useful in some machine learning tasks (the SSD features seem to have good results as a timbre descriptor).

http://ifs.tuwien.ac.at/mir/audiofeatureextraction.html

Python implementation is here: https://github.com/tuwien-musicir/rp_extract