Fix historic mode: slider directly drives LED matrix instead of auto-playback

[Copilot]

Historic mode used autonomous timestamp playback in driver.py — on every Streamlit re-render the UI wrote the same viewing_timestamp back to metadata, causing the driver to endlessly reset to the slider's position and auto-advance independently of user input.

Changes

raspi_code/driver.py

• Main loop: Replaced the auto-playback state machine (starting_timestamp, viewing_idx, historic_timestamps, historic_start_index, speed, ticks) with a single last_historic_timestamp guard. The driver now renders exactly the timestamp the slider writes to metadata — nothing more.
• historic_mode(): Added clear_all_leds() before each render to prevent LED state from bleeding across timestamp transitions. Removed the update_metadata(in_type=LIVE) fallback on missing data, which caused a tight mode-switching loop between the UI and driver when a timestamp had no records.

# Before: auto-advance independent of slider
if db_state.viewing_timestamp != starting_timestamp or speed != db_state.speed:
    viewing_idx = historic_timestamps.index(starting_timestamp)
else:
    viewing_idx += 1          # driver races ahead on its own

# After: slider is sole authority
if current_viewing_timestamp != last_historic_timestamp:
    station_states = historic_mode(station_states, current_viewing_timestamp)
    last_historic_timestamp = current_viewing_timestamp

frontend/app.py

• Removed the "Playback Speed" slider and its speed argument from update_metadata() — meaningless now that the driver no longer auto-advances.

Original prompt

Currently, there's significant issues with historic mode synchronizing with the raspberry pi driver & the streamlit frontend app.py. Carefully redesign the historic timestamp mode so that, when users specify a given timestamp from the slider, the raspberry pi writes to the driver to carefully update the LED matrices.

Created from VS Code.

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.