Add more figures to the summary

[rylev]

This adds more figures to the summary.

Note: I've purposefully haven't worked on finishing the visual presentation.

The figures include:

• all: all test results including insignificant and filtered out results
• filtered: whatever has not been filtered out by a filter
• scenario kinds: the various scenario kinds there are (full, incr-full, incr-unchanged, incr-patched) regardless if they have been filtered.

Also a border (red or green) is applied around any summary element that has more improvements or regressions than unchanged results. This is an attempt to draw attention to summary elements that are "significant".

Questions:

• which summary elements are most useful?
• Is it useful to highlight summary elements that show a "significant" number of improvements or regressions? If so, should it be how it's currently done (if there are more improvement or regression results than unchanged) or by some percentage of the total number of results? If by percentage, which?
• Are we missing any good summary figures?
• Should we try to summarize the summary (i.e., attempt to interpret the results for the viewer so that they do not have to)?

Fixes #935, fixes #188, fixes #953

[Mark-Simulacrum]

Might also fix #953, not sure -- I had forgotten about this when opening that issue.

[rylev]

This fixes #953

I've removed the explicit summaries for the different scenario kinds. We can add those back if we want, but for now that information can be retrieved by explicitly filtering.