Rewrite DiffBind exercise

[AshKernow]

The current DiffBind exercise only uses 2 replicates, not a good example. We should find a different data set with at least 3 reps. One possibility would be the data from this paper:

https://www.ncbi.nlm.nih.gov/pubmed/27980063
MCF10A dataset - https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE72931
MCF7 datasets – Use BRD4 E2 samples https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE55921

There are some nice clear heatmap plots of differentially bound regions in the two cells lines and a third ER+ cell line that we could reproduce as a final outcome of the analysis.