Revamp benchmarks/suite1, as promised on the hackathon. A slight code bomb.
Use an SQLite db to record all timings ever measured. Add capability to generate charts from recorded data. Commandline interface has changed, featuring a way to indicate slices of recorded timings to show/compare/chart.
* tools/dev/benchmarks/suite1/run: Adjust so that it does roughly the same, except, since all timings are now recorded, only run the 1.7.0 tag once instead of N times every week. Todo: Actually ask benchmark.py for charts and show them to the dev list.