[P0] `init_scheduler()` at module level — fires on every gunicorn worker

[ShadowWalkerNC]

Summary

init_scheduler() is called at module import level (bottom of app.py). This means it fires on:

• Every gunicorn worker process (scheduled posts fire N times with N workers)
• Every pytest import (background jobs run during the test suite)
• Every hot reload in development

Impact

Duplicate post publishing in production. Flaky tests. Unpredictable behavior on Railway with multiple workers.

Fix

# Option A — guard with __main__
if __name__ == '__main__':
    init_scheduler()
    app.run()

For multi-worker production, use APScheduler's Postgres jobstore so only one worker claims each job, or run the scheduler as a dedicated worker process via the Procfile.

Effort: ~30 minutes

Part of full audit: #1