Remove C models

[RalfG]

• Replace remaining C model files with .xgboost models
• Use more accurate mass values
• Fix various C compile-time warnings
• Format C and Cython code
• Fix calculating correlations if some peptides could not be processed (e.g., invalid amino acids). Now returns nan for those spectrum correlations
• Clean up setup.py

[RalfG]

Superseded by #260.