Real Player Roster inspired development curves

[nicidob]

This is based on analyzing your real player rosters. The analysis script is here.

Sadly the real test.. "does PER or WS/48 or BPM match correctly" hasn't been performed. I've just been eye-balling the clipping bounds. But things I think are right

• Rookie base stats (based on an average)
• Big/Wing/Big scaling (clustering of players)
• The correct averages for all stats categories (based on player aging data)
• Grouped development (based on feature clustering)