Cool file BALANCE normalization

[claricehong]

I recently discovered this amazing package and am really happy with it so far!!

I was using the development branch of the package so that I can use it with cool files, but I realized that when reading the normalized (balanced) counts, the balanced counts are larger than the actual counts, and the plotted maps look really strange. I looked at the source code and found that the weights are being treated as divisive (counts / (bin1norm * bin2norm)), but I think cooler uses multiplicative weights - at least that's what I understand from the documentation (pasted below, and also my understanding may be completely wrong!).

bias : 1D array, whose shape is the number of bins in ``h5``.
        Vector of bin bias weights to normalize the observed contact map.
        Dropped bins will be assigned the value NaN.
        N[i, j] = O[i, j] * bias[i] * bias[j]

Please let me know if I am just doing something wrong. Thank you for the package and for the continued development as well!