loglik_log_reg_grad: Computes the gradient of the log likelihood of a logistic...
In TomMayo/kmermods: Represents Kmers As Integers Along The Genome

This calculates the prediction for each region, in a non-sliding scheme, calculates the error and returns the update vector for all of the regions.

1 2	loglik_log_reg_grad(kmers_win, params, peaks, win_size, chrom_loc, warp_ = NULL)

`kmers_win`	is a vector of integers of any length representing kmers in a region
`peaks`	is a matrix giving the locations of the peaks on the chromosome, the first column is starts, second is ends, inclusive, indexed from 1
`win_size`	is the length of the sliding window we are using
`chrom_loc`	is the position of the first kmer along the chromosome - this avoids indexing errors when splitting up the data //' @param warp is a vector of length as long as the kmer vector, with the multiplicative weights for how much to warp the entry
`paras`	is a vector of length equal to the total number of kmers