Extend interactivity in make_mapper_pipeline to the scaler parameter

[ulupo]

Currently, the scaler parameter in make_mapper_pipeline represents an arbitrary scikit-learn estimator which will be interpreted as serving to scale the data. In practice, however, this could be improved for the following reasons:

1. The scaling step can always be included as part of filter_func (by feeding a Pipeline for the latter), so this optional argument is little more than a convenience feature, which may cause confusion;
2. Because scaler can only be a scikit-learn estimator, it cannot be edited interactively in plot_interactive_mapper_graph.

To solve 1 and 2 simultaneously, I propose the following:

1. Make scaler be one of a few simple string choices ('minmax', 'standard', 'robust', and the like).
2. Allow the user to toggle between these choices in the interactive widgets produced by plot_interactive_mapper_graph.