metadata for evaluation results

[wsavran]

we should have an open discussion about what quantities of the forecast should appear as meta data in an evaluation result file. this comes from the fact that im working with !~1750 forecasts all having different start times, and it made me wonder how much of the forecast metadata should be associated with the result .json file.

[wsavran]

this could be considered at the forecast level in the form of summary statistics. these would include things like

1. min / max magnitude
2. min / max event count
3. p5, p50, and p95 percentiles of number distribution
4. start / end times