feat: Enable isolated node eval for answer generator nodes (incl. OpenAI Node)

[tholor]

Related Issues

• fixes Enable isolated_node_eval for AnswerGenerator nodes (incl. OpenAI's) #3035

Proposed Changes:

Making the BaseGenerator 's run method more similar to the one from BaseReader by

• adding a param add_isolated_node_eval
• in case the mode is active, taking the same approach as in BaseReader here. The only difference here is that we are not adding the meta data of documents, as the answers are not extracted from any direct docs but are generated. So a direct link here seems impropropriate from my perspective

How did you test it?

• manual test by replacing the reader in our Tutorial 5 by: reader = OpenAIAnswerGenerator(api_key="<api-key>")

Notes for the reviewer

• Let me know if you also agree, that the metadata linking (as done in the BaseReader) doesn't make sense here for the generator
• Just touched the run() method here. Didn't check if run_batch also needs a fix or how the eval mode in batch mode currently works.
• Don't see big side effects possible here as the new if block is quite isolated from anything else

Checklist

• I have read the contributors guidelines and the code of conduct
• I have updated the related issue with new insights and changes
• I added tests that demonstrate the correct behavior of the change
• I've used the conventional commit convention for my PR title
• I documented my code
• I ran pre-commit hooks and fixed any issue

[tstadel]

LGTM!
Just had to fix a small mypy complaint.