New features
- TimeTrains
- Vector output in benchmarks
- plot_legend can now be a python dict
- New reductions : sum, logavg, first_el, last_el, random_el
- Two additional posts in tutorial
- TimeTrain and Constants in docs
Bug fixes
- Better handling of edge cases for out of date benchmark detection
- Try / catch around deepcopying of setup results
- Better defaults for function names in plot_benchmark
- relative_val with intent = same would plot wrong values. Fixed.