Add reproducible PAI public data pipeline

[amit-spatial]

What changed

• adds a browser-free, resumable downloader for FY 2022–23 and FY 2023–24 GP-level PAI scores
• validates portal hierarchy parity and preserves unpublished, malformed, and external-LGD exception records separately
• adds official grade metadata, GP peer ranks, administrative summaries, tests, and reproducible analysis commands
• adds concise scraping, handoff, CoRE Stack, and professor-facing documentation

Why

The public portal does not provide one complete national download, and state-wide pagination can lose rows at tied-score boundaries. The block-level pipeline validates every GP ID against the portal roster before accepting a partition.

Data and privacy

Raw datasets, scrape partitions, source PDFs/spreadsheets, prompts, environment files, and local history remain ignored and are not included in this PR.

Checks

• uvx ruff format --check src tests main.py
• uvx ruff check src tests main.py
• uv run python -m unittest discover -s tests -v — 13 tests passed
• staged privacy scan — clean