A new datatractor CLI

[PeterKraus]

Addresses #10 by implementing

• datatractor beam input_type input_path - install & extract from file
• datatractor probe input_type - return registered Extractors for that input_type
• datatractor yard extractor_id - return extractor definition from registry
• datatractor install extractor_id - install Extractor (optionally into venv)

[ml-evs]

Will dig more into this later, just one comment for now!

[ml-evs]

Couple of minor comments!

[ml-evs]

I think for the CLI, we should try to return valid JSON rather than printing Python dictionaries so it can be piped into other tools, e.g.,

Suggested change

	try:
	ret = func(**vars(args))
	if ret is not None and len(ret) > 0:
	print(ret)
	except RuntimeError as e:
	print(e)
	try:
	ret = func(**vars(args))
	if ret is not None and len(ret) > 0:
	print(json.dumps(ret))
	except RuntimeError as e:
	print(e)

This might need some better error handling though.

[PeterKraus]

Yeah that's true, I forgot a printed python dict is not json.

[ml-evs]

I've added a guarded JSON dump, also when using probe you now get e.g. {"id": "biologic-mpr", "registered_extractors": ["yadg", "galvani"]} back rather than just the list so its valid JSON.

[PeterKraus]

LGTM! (Cannot approve my own PR)