`generate_demo_data` crashes with `BufferError: Local: Queue full`

Bug Description

generate_demo_data management command crashes with BufferError: Local: Queue full when producing simulated events to Kafka.

The _save_past_sim_events method in products/demo/backend/logic/matrix/manager.py writes all events for each simulated person in a tight loop via create_event() → KafkaProducer.produce(), without ever flushing the producer's internal buffer. When the number of events exceeds the librdkafka buffer capacity, the producer raises a BufferError.

How to reproduce

1. Start the dev stack with hogli start
2. The command crashes during demo data generation with:

BufferError: Local: Queue full

Additional context

• The Kafka broker is healthy
• _save_past_sim_events iterates every event and calls create_event() without any periodic flush() call on the Kafka producer
• The producer does poll(0) (non-blocking) after each message, but this only triggers delivery callbacks — it does not wait for the buffer to drain
• A fix would be to periodically flush the producer (e.g. every N events) to let the buffer drain before producing more messages

Debug info

- [ ] PostHog Cloud, region and project ID: [please provide, you can find both at https://app.posthog.com/settings/project-details#variables]
- [ ] PostHog Hobby self-hosted with `docker compose`, version/commit: [please provide]
- [ ] PostHog self-hosted with Kubernetes (deprecated, see [`Sunsetting Kubernetes support`](https://posthog.com/blog/sunsetting-helm-support-posthog)), version/commit: [please provide]

Start in products/demo/backend/logic/matrix/manager.py at _save_past_sim_events, then trace create_event() to KafkaProducer.produce(). Check how the producer currently uses poll(0) and determine where periodic flushing belongs. Run the demo stack with hogli start and confirm generate_demo_data completes without BufferError: Local: Queue full.