When processing documents in chunks, some chunks may contain no entities of interest. Local models (e.g., Gemma3) often respond with polite refusal messages instead of valid JSON, causing the entire extraction job to fail even if other chunks succeeded.
From #132 (@thekitchenscientist):
If the chunks ends up not containing any of the entities of interest, the LLM sends a polite refusal message (as trained to do). This is not valid JSON and lx.extract rejects the whole job even if the rest of the chunks were valid.
Related: #132