[BUG] Horizon SSE stream has no reconnection logic - automation stops silently after network drop

[thisisouvik]

Description

In backend/src/engine/horizonStream.ts, the Horizon SSE stream monitors wallets for incoming payments. There is no exponential backoff or max-retry logic for stream disconnections. If Stellar Horizon drops the connection (common after 30-60s on testnet), the stream silently stops and users' rules stop firing until the backend is restarted.

Steps to Reproduce

1. Start backend with a monitored wallet.
2. Wait for or force the Horizon SSE connection to drop.
3. Send a payment to the wallet.
4. Observe no payment event is processed.

Expected Behavior

The stream should automatically reconnect with exponential backoff (1s, 2s, 4s... max 60s).

Impact

Medium - Rules silently stop executing without any user-visible error or alert.

[nwabaj2017]

@nwabaj2017 has applied to work on this issue as part of the Stellar Wave Program's 8th wave.

Hello maintainer, I'd be delighted to work on this. Please assign

ℹ️ Repo Maintainers: To accept this application, review their application or assign @nwabaj2017 to this issue.