Use Becker et al. 2019 model for Kd and kcat

They have established a very good model from high-throughput experimental data that we should use.

• implement the dynamic programming alignment to match nucleotides
• reuse the weights for the generalized model