Record pilot_id and flight_id in the logged CSV #211

Is your feature request related to a problem? Please describe.
Clustering spans 10 to 20 pilots with multiple flights each, but every CSV row today is anonymous. There is no way to group rows by pilot or by flight, and no join key to any pilot metadata.

Describe the solution you'd like
Add --pilot-id and --flight-id arguments to inference/Data/data_logger.py, written into each row, or into a session sidecar file alongside the CSV. Generate a flight_id when none is given. Update the inference/ scripts that read the CSV to tolerate the new columns.

Acceptance criteria

• Logged CSV contains pilot_id and flight_id columns on every row
• Values are set once per session (config or CLI arg), not per sample
• Missing pilot_id fails loudly at startup rather than writing blanks
• docs/telemetry_schema.md updated if column names/positions change
• Unit test covers a logged file containing both columns

Blocks: #214, #216

Additional context
docs/telemetry_schema.md section 4a explains why this belongs at the logger and not in the plugin or relay: baton's send takes a CxxVector<f32> with no room for a string, and both ids are constant for a whole session.