Standardise reply when question is not answered

[petermr]

The backend gives varying and often opaque answers when it can't answer.
Example:
The assistant's reply wasn't in the format this app expects, so nothing could be shown. Try asking again, or ask in a simpler way.

This should be replaced by:
"The chatbot couldn't find an answer in the book"

[aleenaharoldpeter]

I updated the fallback message in climate_streamlit/llm/parsing.py from:

"The assistant's reply wasn't in the format this app expects, so nothing could be shown. Try asking again, or ask in a simpler way."

to:

"The chatbot couldn't find an answer in the book."

However, the change is not being reflected in the frontend.

While reviewing the file, I also noticed another user-facing fallback message:

"The assistant replied, but none of its paragraphs contained readable answer text (for example empty sections or placeholders). Try asking again, or break the question into parts."

Should this message also be replaced with a simpler book-focused response such as:

"The chatbot couldn't find an answer in the book."

At the moment, I updated the exact message that was requested, but the frontend still appears to be displaying the original text.