[JSON output] Make export module for meal completion interface about batch5_phenix_happymeal HOT 5 OPEN

The function optimize_baskets in https://github.com/dataforgoodfr/batch5_phenix_happymeal/blob/master/meal_balancer/algos_optimisation/milp_cvxpy_glpk.py already returns a json with the two camamberts (distribution of items allocated to balanced baskets and distribution of remaining items) and the keys "nb_balanced_meals", "nb_remaining_items", "pct_weight_allocated_items", "pct_weight_remaining_items" (see example https://data-for-good.slack.com/archives/CDJF42F28/p1548275012014100)

from batch5_phenix_happymeal.

(Yassi is working with this json as his input)

from batch5_phenix_happymeal.

To add the final balance_adjuster, we should

• output from optimize_baskets a dataframe with all necessary info for balance_adjuster and the results dictionary for the two camemberts (as in https://data-for-good.slack.com/archives/CDJF42F28/p1548275012014100)

• Then balance_adjuster can take the info in the dataframe and update the results dictionary with the data for the additional plot.

Note: we should remove from meal_balancer.algos_optimisation.postprocess_optimised_solution_for_ui the lines

with open(results_filename, 'w') as f:
    json.dump(results, f)

and output the json after balance_adjuster instead

from batch5_phenix_happymeal.

@frgfm je vais ouvrir une issue à part pour la partie 'se coordonner sur la sortie de optimize_baskets/entrée de balance_adjuster'

from batch5_phenix_happymeal.

Our code should return a stringified json, instead of writing the json to disk

from batch5_phenix_happymeal.