Simple output for plot creation

[Drup]

The output of benchmark_js are not terribly useful (they expose you to the full linear regression, which is ... a bit overkill).

However, the inner workings of bechamel are complicated enough that I couldn't figure out to make a "simple" output to make plots with confidence intervals.

In the ended, I made a jq script to output something of the form :

[
  {
    "name": "map_par_full",
    "nb": 100,
    "time": 139578.8546148588,
    "rsquare": 0.9743917702585394
  },
  {
    "name": "map_par_full",
    "nb": 1000,
    "time": 95153.07190147786,
    "rsquare": 0.9786795220119054
  },
...
]

With is then easy to use in mathplotlib.
I would be nice to have a default output like that, possible using csv directly instead.

Here is the jq script I used:

[.series.[] |
(.name | capture("tree bench (?<name>[a-z_]+) (?<run>[0-9]+)")) as $scan |
($scan.run | tonumber) as $nb |
{ name : $scan.name,
  nb: $nb,
  time: (.result.estimate | . / $nb),
  rsquare: .result.r_square}
]