Add benchmark script, fix regex, remove N806

[jonashaag]

Adds a script that outputs a table like

Model	Baseline Size	Our Size	Size Reduction	Baseline Loading Time	Our Loading Time	Slowdown
sklearn RandomForestRegressor	181.2 MiB	24.1 MiB	7.5 x	0.0 s	0.1 s	4.2 x
lightgbm LGBMRegressor	0.3 MiB	0.1 MiB	3.1 x	0.0 s	0.0 s	10.7 x

Feel free to change this however you prefer. Not sure if I will find more time to work on this.

[github-actions]

(benchmark 4245116652 / attempt 1)
Base results / Our results / Change

Model	Size	Dump Time	Load Time
RandomForestRegressor	77.2 MiB / 12.5 MiB / 6.16 x	0.07 s / 0.11 s / 1.53 x	0.08 s / 0.09 s / 1.08 x
LGBMRegressor gbdt	0.3 MiB / 0.1 MiB / 2.64 x	0.00 s / 0.02 s / 3.44 x	0.00 s / 0.02 s / 12.02 x
LGBMRegressor rf	1.6 MiB / 0.5 MiB / 3.30 x	0.05 s / 0.08 s / 1.80 x	0.01 s / 0.08 s / 13.99 x

[pavelzw]

Closes #6

[pavelzw]

It could be interesting to integrate benchmark-action. But I think for now this should be enough...