No retry/backoff for transient Soroban RPC failures

[DeFiVC]

What

getLatestLedger has no try/catch; a network error propagates to the outer catch which logs and sleeps for a fixed pollIntervalMs (5s). If the RPC is down for hours, the indexer hammers it every 5 seconds with no exponential backoff. Similarly, fetchContractEvents catches RPC errors and returns an empty array — no retry within the same cycle.

Why

A single failed RPC call wastes the entire cycle for that contract. No circuit breaker exists to pause requests during prolonged outages.

Scope

• Add exponential backoff for RPC failures
• Implement a circuit breaker pattern
• Retry failed RPC calls within the same cycle

Acceptance Criteria

• RPC failures trigger exponential backoff
• Prolonged outages don't cause rapid retry loops
• Failed RPC calls are retried within the same cycle

Technical Context

• File: src/indexer/event-listener.ts, lines 115-119, 128-131
• Consider using the cockatiel library (already used in chainlearn-api)