huggingface-cli upload - Validate README.md before file hashing

[hlky]

For large datasets file hashing can take some time.

Dataset Card validation happens after file hashing.

"It's better to fail early than to fail after all the files have been uploaded hashed."

[Wauplin]

Hi @hlky, thanks for noticing this! It would indeed be much better to validate the yaml before.

For reference, the files are hashed when the CommitOperationAdd objects are defined here. You can either call the /validate-yaml endpoint between the moment we have the list of paths and the moment we create the CommitOperationAdd objects. Otherwise, it should be possible to compute the upload_info attribute (which is when the hash is computed) only on demand (lazy-compute) instead of when initializing the object (see here).

Would you like to try to work on a PR to improve this?

[hlky]

Thanks for the reference, I've created a PR #2452