fix: guard _runFocus against disposed controller (forward() after dispose)

[thinklanguages]

Problem

AnimatedFocusLightState._runFocus() is scheduled asynchronously — Future.delayed(Duration.zero, _runFocus) in initState, plus an await widget.beforeFocus(...) — and then calls _controller.forward() with no mounted check.

If the widget is disposed during that async gap (the host overlay is removed via removeOverlayEntry() / a route pop while the tutorial is mid-presentation), forward() runs on a disposed AnimationController. _ticker is null, and because assert(_ticker != null) inside AnimationController.stop() is stripped in release builds, it throws an uncaught Null check operator used on a null value (_TypeError) instead of a debug assertion.

We hit this in production (release) on a coach mark shown on a screen the user navigated away from before the focus animation started. In debug it surfaces as:

AnimationController.forward() called after AnimationController.dispose()
'package:flutter/src/animation/animation_controller.dart': Failed assertion: '_ticker != null'
  AnimatedFocusLightState._runFocus (animated_focus_light.dart)

Fix

Add a mounted guard immediately before forward() in _runFocus:

  safeSetState(() { ... });

+ if (!mounted) return;

  await _controller.forward();
  _isAnimating = false;

safeSetState above is already mounted-guarded, so this is the only remaining unguarded controller access on that path. Minimal, no behavior change when the widget is still mounted.

(The same pattern exists on the develop branch's restructured _runFocus; happy to mirror it there if you prefer.)