fix: type of temperature param and adjust defaults for OpenAIAnswerGenerator

[tholor]

Proposed Changes:

• Small fix that changes the type of temperature param from int to float. Their API allows floats and this was causing problems in the YAML schema validation.
• Also adjusting the default values as tests showed that with max_tokens = 7 quite many answers get cut off in the middle. GPT-3 tends to be a bit more verbose in it's answers
• Increasing the default temperature slightly, as tests showed that we otherwise get often multiple answers generated that are exactly the same as the "top 1"

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

[Timoeller]

Arr, I was playing with it as well today and have similar code changes.

About temperatur=0.2: when using it in our pipeline.eval function on AdversarialQA dataset with topk retriever=5 amd topk generator = 5 I mostly got 5 identical answers.

I will open a PR later as well with some more little tweaks.

[vblagoje]

LGTM @tholor