[Feature] Support delayed event delegation for legacy JWT request

[fkwp]

If, e.g., the homeserver is restarted due to k8s rescaling the CS API is offline for a short amount of time. If in this period an Element Call client disconnects (closing browser tab, network issues, etc.) the belonging MatrixRTC membership is not cleaned up. This resolves once the delayed event with the leave event is sent. Typically the delayed event timeout is in the order of 30-90 seconds, which is perceived as a long period of time.

This PR

• adds the ability to add the delayed event delegation information also to the legacy JWT request, which reduces the emission of the delayed leave event to a minimum.
• Increases the maximumNetworkErrorRetryCount to NOT fail during (the increased) timeouts.

Note: The deployed MatrixRTC stack requires a MatrixRTC Authorization Service (lk-jwt-service) to support the delegation feature, which is currently a PR.

[element-hq-sentry-production]

🔍 Existing Issues For Review

Your pull request is modifying functions with the following pre-existing issues:

📄 File: src/livekit/openIDSFU.ts

Function	Unhandled Issue
getSFUConfigWithOpenID	Error: Something went wrong /room/ Event Count: 1 Affected Users: 0
getSFUConfigWithOpenID	Error: Etwas ist schief gelaufen /room/ Event Count: 1 Affected Users: 0

---

_{Did you find this useful? React with a 👍 or 👎}

[BillCarsonFr]

Looks like this is doing a bit more that delegation? The js-sdk thing about maximumNetworkErrorRetryCount?

[fkwp]

Looks like this is doing a bit more that delegation? The js-sdk thing about maximumNetworkErrorRetryCount?

fair -- however, its very related.

[element-hq-sentry-production]

Issues attributed to commits in this pull request

This pull request was merged and Sentry observed the following issues:

• ‼️ Error: Something went wrong in production

• ‼️ Error: Something went wrong in call.element.io

_{Did you find this useful? React with a 👍 or 👎}