Bootstrap AppleWatchAI: implement all empty source modules, tests, and fix config

[Copilot]

The project had a complete scaffolding (Poetry, DuckDB, Pandera, Ruff, Dynaconf all wired up) but every file under src/awai/ was empty — zero implementation code, no tests, and several config issues that blocked even installing the package.

Core implementation

• utils/xml_parser.py — loads export.xml via xmltodict; extracts records, workouts, and activity summaries into DataFrames. Adds _ensure_list to handle xmltodict returning a dict (not a list) when only one XML child element is present.
• utils/data_cleaner.py — camelCase → snake_case column renaming, date parsing + duration calculation, metadata column dropping, value normalisation (non-numeric → 1.0 for presence indicators), HK-prefix stripping, per-type filtering, daily & monthly aggregation.
• models/schemas.py — Pandera RecordsSchema validating the cleaned DataFrame (types, YYYY-MM-DD date format, non-negative values).
• tasks/load_data.py — idempotent XML → DuckDB loader; existence check scoped to main schema to avoid false positives from other databases.
• tasks/prepare_data.py — full pipeline: DuckDB read → clean → filter by type → daily & monthly aggregation.
• entrypoint.py — Fire-based CLI exposing two subcommands:

awai load    --xml_path=data/export.xml --db_path=data/health_data.duckdb
awai prepare --db_path=data/health_data.duckdb

Tests (28 passing)

• tests/test_xml_parser.py — XML load, single-element edge case, all three extract functions.
• tests/test_data_cleaner.py — camel_to_snake parametrised cases, clean_records column shape/dtype/drop behaviour, filter_record_types, daily & monthly aggregation sums.

Config fixes

• pyproject.toml — yanked poetry-core==1.8.2 → >=1.9.0; added [tool.poetry.scripts] entry and [tool.pytest.ini_options].
• .pre-commit-config.yaml — files glob was ^src/.*\.pyi$ (stub files only); corrected to \.pyi?$ to cover .py files.
• settings.toml — populated with data_dir, xml_file_name, db_name, log_level defaults.
• init_setup.sh — removed unused npm/nodemon steps.
• README.md — rewritten with accurate project structure, setup, and CLI usage.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.