feat(price_pusher): exponential backoff for Hermes reconnect

[dhruvja]

Summary

The Hermes onerror handler in PythPriceListener retried with a flat 5-second delay and fire-and-forget void this.startListening(), which has two problems:

• Every pusher instance globally hammers a flapping Hermes endpoint every 5 seconds with no backoff.
• If the reconnect itself (getPriceUpdatesStream) rejects, it becomes an unhandled rejection.

This extracts the reconnect into a reconnectWithBackoff method that doubles its delay (1s → 30s cap) on each failed attempt, routes errors through the pino logger, and catches reconnect-time failures so they feed the next iteration instead of going unhandled. The existing health
check (which throws after 30s of no updates) still bounds the worst case.

How has this been tested?

• Current tests cover my changes
• Added new tests
• Manually tested the code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

2 Skipped Deployments

Project	Deployment	Actions	Updated (UTC)
api-reference	Skipped		Apr 22, 2026 7:42am
proposals	Skipped		Apr 22, 2026 7:42am

[vercel]

@dhruvja is attempting to deploy a commit to the Pyth Network Team on Vercel.

A member of the Team first needs to authorize it.