Build some performance tests

[benbasson]

To assess whether changes made to address #38 and #39 are beneficial / find out which method is optimal, we will need some rudimentary performance metrics. I propose building a small page that automatically runs tests on tagger widgets with different amounts of data so that we have some idea of current performance and can tell if we're improving it or not.

As most of our time goes in either building the widget (because we need to know what values are selected) or building the suggestions list, I'm planning to start with those as the major focus and use the same configuration (a multi-select) with differing numbers of values.