fix: add missing commas in LLM prompts to ensure valid JSON output

[Himanshuwagh]

Title: Fix JSON parsing errors by adding missing commas in LLM prompts

Summary

This PR fixes intermittent KeyError and JSON parsing failures by ensuring all LLM prompts in page_index.py request a valid JSON structure.

The Problem

Several prompts were missing a comma between the "thinking" key and the subsequent result key. This caused the model to output invalid JSON syntax, which resulted in the following error:

KeyError: 'toc_detected' (and similar for 'completed', 'page_index_given_in_toc')

Changes Made

Added trailing commas to the "thinking" lines in the following functions within pageindex/page_index.py:

• check_title_appearance (Line 34)
• check_title_appearance_in_start (Line 62)
• toc_detector_single_page (Line 112)
• check_if_toc_extraction_is_complete (Line 132)
• check_if_toc_transformation_is_complete (Line 150)
• detect_page_index (Line 213)

Testing

• Verified that the prompts now strictly adhere to standard JSON formatting.
• This fix resolves the KeyError encountered when the LLM follows the provided format literally.