Add distinction between two unallowed cases in name check "eval_"

[J4nn1K]

What this does

Title	Label
Fixes error in sanity_check_dataset_name()	🐛 Bug

There was a bug within the sanity_check_dataset_name() function. When the control_robot.py script was executed for evaluation and the repo-id did not start with "eval_" it was raising a wrong ValueError.

This caused some confusion because it was unclear that the prefix "eval_" is needed.

How it was tested

• Added distinction between two possible error cases and checked that evaluation works with policy act_so100_real on our own dataset.

How to checkout & try? (for the reviewer)

Examples:
This should raise the right error message:
(Taken from the SO-100 Guide)

python lerobot/scripts/control_robot.py record \
  --robot-path lerobot/configs/robot/so100.yaml \
  --fps 30 \
  --root data \
  --repo-id ${HF_USER}/act_so100_test \
  --tags so100 tutorial eval \
  --warmup-time-s 5 \
  --episode-time-s 40 \
  --reset-time-s 10 \
  --num-episodes 10 \
  -p outputs/train/act_so100_test/checkpoints/last/pretrained_model