fix: initialize form rendering for fillable PDFs (#240)

[rstrahan]

Summary

Fixes #240 — Fillable PDF form fields (text inputs, checkboxes, radio buttons, dropdowns) were not rendered in page images, causing empty extraction results.

Root Cause

pypdfium2's page.render(may_draw_forms=True) (the default) requires PdfDocument.init_forms() to be called first to initialize the form rendering engine. Without this call, AcroForm content is silently omitted from rendered images — so OCR/extraction only sees the blank form template.

Changes

• lib/idp_common_pkg/idp_common/ocr/service.py — Added pdf_document.init_forms() after pdfium.PdfDocument(file_content) in process_document() (Pattern 2 pipeline OCR)
• patterns/unified/src/bda_processresults_function/index.py — Added pdf_document.init_forms() after pdfium.PdfDocument(pdf_content) in create_pdf_page_images() (Pattern 1 BDA thumbnails)
• lib/idp_common_pkg/tests/unit/ocr/test_ocr_service.py — Added regression test test_process_document_calls_init_forms_for_fillable_pdfs
• CHANGELOG.md — Added fix entry under [Unreleased]

Risk Assessment

Very low risk — init_forms() is a no-op for non-fillable PDFs. The may_draw_forms=True default was already set; this fix simply enables the feature that was intended to work.

Testing

• All 666 unit tests pass (make test-cicd)
• Ruff lint clean on all changed files
• Verified fix works with fillable VA Form 21-22a from the issue report