Speed up correct_k

[Carter12s]

Profiling the chemdose_ph_chain examples from the vignettes, correct_k was noticed to be slow:

Applying optimizations contained within:

correct_k goes from 1980ms to 90ms (-95.4% faster)

Overall example goes from 7.74s to 5.67s (-27% faster).

These changes and gains are independent of: #7

[Carter12s]

This this MR with #7 yields:

7.7s -> 2.2s = -68% faster

[sierrajohnson]

The change makes sense, but all the k values in correct_k should be written as 10^x (industry standard) - you can see these in the raw data file. Also, let's keep discons and reference it in the comments of correct_k so that package users can still see where the data come from.