ENH add simple example_config.yml to get started

[mathurinm]

We can commit a simple config with a small dataset, few objectives, some fast solvers (sklearn, celer, cd ?) and add a snippet in the Readme so that people can get here, clone the repo, and get a benchmark in 1 min

May @Klopfe @QB3 or @tbng would have the time to do it ?

[Binh Nguyen]

You mean like this @mathurinm :

# example_config.yml
dataset: # dataset to run benchmark on
  - leukemia
  - simulated

force-solver: # list of example solvers to do benchmark
  - celer
  - cd
  - sklearn

# specify more than 1 repeated runs to have error-bar on the plot
n-repetitions: 1

and add to README:

# Quick Start

To perform a quick test benchmark, one can run:

`benchopt run --config example_config.yml`

This config file will perform benchmarks of `celer`, `cd` (coordinate-descent), and `sklearn` implementation of the sparse solver on 2 dataset: leukemia and toy (simuldated dataset)

Or of course to quick test the library with good plots one can reduce to much smaller dataset and get faster results

[mathurinm]

Yes exactly like this ! Can you send a PR ?

Once benchopt/benchopt#362 we can improve by creating a small simulated dataset for this purpose !

[mathurinm]

Solved in #82