Create real PDF signature fields

[nonprofittechy]

Summary

Fixes FormFyxer PDF export so FieldType.SIGNATURE fields are emitted as real AcroForm signature fields (/FT /Sig) instead of text fields.

Also declares the uv dev dependency group expected by the shared SuffolkLITLab/ALActions/pythontests workflow so CI can install mypy, pytest, and FormFyxer runtime imports before running checks.

Root cause

ReportLab does not expose a signature widget API, so FormFyxer created signature placeholders with form.textfield(). That preserved placement, but external PDF editors correctly interpreted the result as a text field.

The CI failure happened because the shared action runs uv sync --group dev when [dependency-groups] exists and then calls uv run mypy; FormFyxer did not previously declare uv-managed dev dependencies.

Changes

• Normalize generated signature placeholders in the temporary field PDF before copying fields into the destination PDF.
• Set AcroForm /SigFlags when signature fields are created.
• Preserve signature width/height when existing PDF fields are read back into FormField objects.
• Add an end-to-end regression test that writes a PDF with both signature and text fields, then verifies the raw PDF field types.
• Add a dev dependency group for the shared uv-based CI action.

Validation

• uv sync --group dev
• export PYTHONPATH=$PYTHONPATH:$PWD; uv run mypy . --exclude '^build/' --explicit-package-bases
• uv tool run bandit -r . --exclude './scripts,./venv,./.venv,./build' --severity-level=high
• export PYTHONPATH=$PYTHONPATH:$PWD; uv run pytest -q
• python -m py_compile formfyxer/pdf_wrangling.py formfyxer/tests/test_pdf_labeling_rules.py
• git diff --check

GitHub Actions passed on commit 0afd913.