Feature/benchmarking scripts

[ffelten]

Relates to #12

Wishlist:

• [x] Explicit seeds
• [x] Explicit wandb entity
• [x] Shell script: algo-name, env-id, gamma, wandb entity, timesteps, seed
• [x] standardize algorithms init and train to take parameters above
• [x] openrlbenchmark stored run and extract plot

Example plot generated using this query on openrlbenchmark CLI:

 python -m openrlbenchmark.rlops \
    --filters '?we=openrlbenchmark&wpn=MORL-Baselines&ceik=env_id&cen=algo&metric=eval/eum' \
        'PCN' \
        'MultiPolicy MO Q-Learning' \
        'Envelope' \
    --env-ids deep-sea-treasure-v0 deep-sea-treasure-concave-v0 \
    --ncols 2 \
    --ncols-legend 1 \
    --output-filename compare.png \
    --report \
    --scan-history

[ffelten]

@LucasAlegre very much a WIP but I think it might look like this in the end if we stick to the default hyperparameters. What do you think?