Select propagation implemtation

[tdudgeon]

The general idea of selection propagation is described in #11

We are now ready to start implementing this as the scatter plot is now able to handle selections. It should provide a good model for other visual cells.

At the chart/javascript level it now supports:

As inputs:

1. filtering: This is a set of UUIDs and ONLY those datapoints are displayed in the plot.
2. marking: This is a set of UUIDs and those datapoints are highlighted in the plot.

As outputs:

1. selection: an array of the UUIDs of the brushed elements (those in the region that has been dragged out in the plot)
2. selection+marked: an array of the UUIDs of the brushed elements that were ALSO marked in the input.

With this we should now be able to connect these outputs/inputs. Scatterplot would need to connect to scatterplot until we have other examples. What we now need is the cell endpoints for this (and then the ability to connect them). Scatterplot would thus have 3 inputs: data (already present), filter, mark and 2 outputs: selected, selected+marked. We should distinguish these with different colours and shapes (maybe filter and mark the same colour (blue?) but different shape?).

Typical scenario could be:

Data cell providing data.
Scatterplot1 and Scatterplot2 both with input from the data cell output.
Scatterplot2 has its mark (or filter) from the selection of Scatterplot1.