Refactor schedule mapping

[MTrab]

Summary:

• move schedule parsing to a slots-first model so both protocol 0 d arrays and protocol 1 slots keep every run in schedules["slots"] while Schedule, ScheduleInfo and DeviceHandler base progress/next calculations on those slots instead of primary/secondary.
• expose the richer slot metadata (duration_extended/end/source/party_mode/one_time flags) on the public device APIs and keep the raw cfg/dat dumps aligned.
• refresh documentation (README.md, DATA_MAPPING.md) and scripts/dump_mapping.py so the reference fixtures describe the new mapping and the script tolerates multiple payload entries per fixture.

Changes:

• refactored ScheduleParser, Schedule and DeviceHandler to drop primary/secondary and consume only slots, plus updated schedules metadata keys.
• updated docs to explain the slot-first approach, including the new metadata fields and source tags.
• made scripts/dump_mapping.py decode fixtures with multiple payloads so the visual dump mirrors what DeviceHandler exposes.

Testing:

• bash scripts/prepare_test_fixtures.sh
• source .venv/bin/activate && pytest tests/test_device_decode.py -q
• source .venv/bin/activate && python scripts/dump_mapping.py