Resolve "Analysis of playlist corpus"

Closes #3 (closed)

Offers insights for sampling. Performs EDA on complete playlist data set. Steps include:

• Popularity aggregations of songs and tracks
• Adding popularity stats per playlist
• Overview of most popular items and cross-check with official stats
• Density plots on continuous features (including new popularity feature)
• New ms_mean feature to represent playlist length on a track basis
• Bi-Variate Analysis (Correlations, p-Values)
• Analysis of text information in header and description (including emojis, tf-idf scores, normal frequencies and bigrams)