fix(queue): enforce hard backpressure on bounded queues

[Babigdk]

Summary

Implements bounded-memory backpressure controls for stream ingestion queues.

The existing queue already enforced a capacity limit, but non-critical packets were dropped immediately when the queue became full. This change applies hard producer-side backpressure so upstream ingestion blocks until downstream consumers free capacity, preventing uncontrolled memory growth during sink lag events.

Related Issue

Closes #487

Changes Made

• Verified bounded queue capacity is enforced with a default maximum size of 1000
• Updated enqueue logic to block producers when queue capacity is exhausted
• Preserved saturation-based slowdown behavior
• Preserved metric packet shedding under high saturation conditions
• Improved protection against memory exhaustion caused by slow downstream writers

Backpressure Behavior

Before:

• Critical packets waited
• Non-critical packets were dropped when queue became full

After:

• Queue remains strictly bounded
• Producers block when capacity is exhausted
• Upstream readers naturally slow until consumers drain the queue
• Metric packets may still be dropped under configured saturation thresholds

Acceptance Criteria

• Internal queue storage remains bounded
• Maximum capacity enforced at 1000 items
• Backpressure blocks upstream ingestion when capacity is reached
• Prevents unbounded memory growth during downstream lag

Testing

Validation performed via code review of queue flow:

• Capacity enforcement remains active
• Producer blocking path uses queue.put(...)
• Existing saturation metrics remain functional

Notes for Reviewers

The repository issue references a Python implementation using janus.Queue(maxsize=1000), but the current codebase implements equivalent functionality in TypeScript via AsyncBoundedQueue. This PR applies the requested bounded-memory backpressure behavior to the TypeScript implementation.

[drips-wave]

@Babigdk Great news! 🎉 Based on an automated assessment of this PR, the linked Wave issue(s) no longer count against your application limits.

You can now already apply to more issues while waiting for a review of this PR. Keep up the great work! 🚀

Learn more about application limits