Improve Documentation When Training With --config_path

[GrmanRodriguez]

What this does

Fixes #888

The documentation (examples/4_train_policy_with_script.md) suggests that you can resume training with the following command:

python lerobot/scripts/train.py \
    --config_path=outputs/train/run_resumption/checkpoints/last/pretrained_model/ \
    --resume=true

The code however expects the path to train_config.json specifically.

This PR updates the documentation to reflect this.

How it was tested

• Running python lerobot/scripts/train.py --config_path=outputs/train/run_resumption/checkpoints/last/pretrained_model/ --resume=true does not work
• Running python lerobot/scripts/train.py --config_path=outputs/train/run_resumption/checkpoints/last/pretrained_model/train_config.json --resume=true does not work

How to checkout & try? (for the reviewer)

Try the two commands above.

[imstevenpmwork]

This LGTM because I couldn't find where we consider the case when this argument is a directory and not a file. But curious to know what @aliberts thinks