Flags and selects in LLM eval

[kasnerz]

Currently, we can only get additional flags and options from human evaluation campaign.

We should extend this possibility also to LLM eval campaigns.

The best way to do it seems to be continuing the chat, appending the extra questions after the annotations are generated (not to confuse the model when generating the annotations).