Horizon Endpoint Availability Prober

[Kingsman-99]

Description

Before the SDK makes any Horizon API call, it should proactively verify that the configured Horizon endpoint is reachable and returning a healthy response. This is especially important at startup and after network configuration changes. A lightweight availability prober should periodically hit the Horizon root endpoint (/), parse the history_latest_ledger field, and mark the endpoint as degraded or unavailable when the response is stale or the call fails.

Technical Context

Add src/horizonProber.ts. Calls @stellar/stellar-sdk Server.fetchRoot() to obtain HorizonApi.RootResponse. Tracks history_latest_ledger and compares it against the expected cadence (new ledger every ~5 s). Integrates with src/health.ts and src/healthDashboard.ts. Emits horizonEndpointDegraded and horizonEndpointRecovered via src/events.ts. Configured as a step in src/loadBalancer.ts health checks.

Acceptance Criteria

• HorizonProber.probe(endpointUrl): Promise<ProbeResult> returns { reachable, latencyMs, latestLedger, isStale } within timeoutMs
• isStale is true when history_latest_ledger has not advanced in stalenessThresholdMs (configurable, default 30 000 ms)
• Emits horizonEndpointDegraded when two consecutive probes fail or return stale results
• Emits horizonEndpointRecovered on the first successful non-stale probe after a degraded period
• src/healthDashboard.ts includes the prober's last result in the dashboard snapshot
• All CI checks (npm test, npm run build, ESLint/TypeScript) pass and the branch has no merge conflicts

---