fix(web): recover in place from fatal hls.js errors

[ik-make]

Summary

hls.js flags many recoverable startup/stall/network conditions as fatal, and its documented contract is for the host to attempt recovery in place (hls.startLoad() for network errors, hls.recoverMediaError() for media errors) rather than destroy the player. See hls.js error recovery.

Previously the hlsError handler surfaced every fatal error straight to the plugin as a terminal MEDIA_ERR_NETWORK PlatformException. Consumers (MessengerX daily content, watch, and discovery players) each had to compensate by fully tearing down and re-creating the VideoPlayerController to recover — a heavy, duplicated workaround that also produced a brief error/black-frame flash.

This PR makes the fork perform hls.js's built-in recovery in place:

• Adds startLoad() / recoverMediaError() bindings to HlsExtension (lib/hls.dart).
• On a fatal hlsError, networkError -> startLoad(), mediaError -> recoverMediaError(), bounded by a small attempt budget (_maxHlsRecoveryAttempts = 3).
• The budget resets once playback resumes (onPlaying), so a later unrelated error gets a fresh set of attempts.
• Unrecoverable errors, or exhausting the budget, still surface the same terminal PlatformException as before (no behavior change for truly fatal cases).

No public API changes beyond the two additive Hls bindings.

Test plan

• dart analyze lib is clean (verified locally).
• Chrome web: HLS stream that emits a transient fatal network error self-heals without a visible error state.
• Truly unplayable source still surfaces the terminal error after the bounded attempts.
• Consumer app (MessengerX daily content, deep-link/direct entry) plays on first load without the Retry flash.

Made with Cursor

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

Want higher recall? High effort reviews run extra passes and find more bugs. A team admin can switch effort levels in the Cursor dashboard.

^{Reviewed by Cursor Bugbot for commit df77369. Configure here.}

[cursor]

HLS recovery after dispose

Medium Severity

Fatal hlsError callbacks stay registered on _hls after dispose() (only stopLoad() runs and video subscriptions cancel). A late fatal error can still enter _handleFatalHlsError and call startLoad() or recoverMediaError(), restarting HLS work on a disposed player.

Additional Locations (1)

• lib/src/video_player.dart#L354-L365

 

^{Reviewed by Cursor Bugbot for commit df77369. Configure here.}