Pushing the interactive plots feature to master

[AvantiShri]

Plots are demonstrated in https://github.com/kundajelab/tfmodisco/blob/3a0360f3ed0a65e48eda07d877478e306a835e08/examples/simulated_TAL_GATA_deeplearning/InteractiveViz_TF_MoDISco_TAL_GATA.ipynb. Seqlets comprising a motif are visualized in a tsne plot, and the user can select a subset of the seqlets (by dragging a rectangle around them on the plot) to aggregate and visualize on the fly. Good for dissecting heterogeneity within a motif. This is what it looks like:

"track_names_and_signs" refers to a list of the track names that are to be aggregated to render the middle row of the interactive plot; the sign indicates what that track will be multiplied by (I'm expecting values of either -1 or 1) prior to doing the aggregation.

On my laptop, I need to run "%matplotlib notebook" twice in order to get the plots to work, for some reason.
Note: I switched from the LassoSelector to the RectangularSelector as the latter is probably easier on the backend.