docs: add evidence-driven RTDS troubleshooting guide

[osr21]

Summary

Adds an evidence-driven RTDS troubleshooting guide organized around:

symptom → evidence → workaround → requested platform fix

The guide:

• provides a reusable production incident-report template;
• distinguishes WebSocket transport liveness from subscribed data-plane liveness;
• documents the silent-open stream failure reported in WebSocket data stream stops after some time #26;
• recommends valid-payload watchdogs and hard reconnects instead of ping-only monitoring or repeated subscriptions;
• includes a minimal TypeScript watchdog pattern;
• lists the environment and raw-frame evidence maintainers need for diagnosis;
• adds a README link so builders can find the guide.

No runtime behavior is changed.

Why

Issue reports are substantially easier to act on when they include exact subscription frames, UTC timelines, raw data-frame evidence, topic-level last-payload timestamps, and a tested workaround. The current client reconnects on socket close/error, but production users also need guidance for cases where a socket remains open while subscribed data stops.

Related: #26, #31, #32, #38.

---

Note

Low Risk
Documentation-only: a troubleshooting guide and README link with no client or server behavior changes.

Overview
Adds a production troubleshooting guide so RTDS issues can be reported as symptom → evidence → workaround → requested platform fix, and links it from the README.

The new TROUBLESHOOTING.md covers a report template (subscription frames, UTC timelines, last valid payload vs ping/pong), how to tell transport liveness from data-plane stalls (including issue #26), and a minimal payload-staleness watchdog that hard-reconnects instead of re-subscribing on a frozen socket. No runtime code changes.

^{Reviewed by Cursor Bugbot for commit 3c09318. Bugbot is set up for automated code reviews on this repo. Configure here.}