If it hits a timeout but the content is still there, it will now return the content (without the error message) to ensure something small like this doesn't get synthesized.
That's the thing about sytnthesizing and I feel one of the strengths of this system, is the LLM doesn't put too much of it's own "creativity" (which usually comes with hallucination) on to the top of the findings!
TL;DR: Fixed.
The Rush University website takes a long time to load (check it out), and the script recognized the article had loaded but something on the website was causing it to hang (waiting for networkidle status) so it terminated the parsing early and worked with the content it had.
So in a nutshell, it still parsed but with a warning/error that this happened.
I'm optimizing this now to exclude the wording from the report, or make a note.
Leslie, I typed your symptoms into the thing up here, and it says you could have network connectivity problems.