Thanks for this question and for using the Forum. I moved this topic to the Support category, because it looks like questions for Rubin staff - and I think I can answer your questions.
Yes, the flux calibrations that Rubin applies do work well for extended sources.
Yes, when cut-outs are made from calibrated images (e.g., calexp or deepCoadd image types), the flux calibration has been applied to the pixel values.
Yes, all calibrated image and catalog data products will already have flux units of nJy. Users do not need to do that step.
When the first real data is released by Rubin Observatory as part of Data Preview 1, the accompanying documentation about flux calibration that you’re looking for will be released.
I’m pretty sure this answers your questions so I’m going to mark this post as the solution for the topic, but please don’t hesitate to reply or open a new topic for new questions at any time.