[unjica]

Fixes #843.

• @telemetry-tracker/core 1.5.2: new flush() (sends the queue, awaits in-flight batch requests, never rejects) and pendingEventCount(). Batching timer unchanged.
• @telemetry-tracker/node 1.4.2: exports async flush(); registers a beforeExit handler that flushes only when events are queued (no exit loop), plus a best-effort exit handler that warns when events are still queued (e.g. after process.exit()).
• Tests: unit tests for handlers and core flush; spawned-process tests against a local mock ingest: the issue's job.mjs delivers 2/2 with no sleep, await flush(); process.exit(0) delivers, long-running process still batches on the interval.
• Docs: /docs/node short-lived jobs section, /docs/alerts heartbeat example and the missing-event alerts page use await flush(), each noting it requires Node SDK >= 1.4.2.
• Not published to npm: core 1.5.2 and node 1.4.2 need publishing (core first).