Bug: page_indices in extraction results always starts at [0] instead of preserving document position

[mmiakashs]

Problem

The page_indices field in extraction service output (split_document.page_indices) incorrectly always starts at [0, 1, 2, ...] for each section, regardless of which pages from the original document packet the section contains.

Current Behavior

• Section with pages [5, 6, 7] outputs page_indices: [0, 1, 2] ❌
• Section with pages [10, 11, 12] outputs page_indices: [0, 1, 2] ❌

Expected Behavior

• Section with pages [5, 6, 7] should output page_indices: [4, 5, 6] ✓
• Section with pages [10, 11, 12] should output page_indices: [9, 10, 11] ✓