feat: YouTube-style tap gestures (double-tap seek zones on mobile, double-click fullscreen on desktop)

[Ortes]

Stacked on #953 — this branch includes #953's commits (the keyboard seek indicator). Only the last 2 commits are new here; it will be rebased once #953 lands.
Supersedes #932.

What

YouTube-style tap gestures for the Material controls:

Platform	Gesture	Behavior
Mobile (MaterialControls)	Double-tap left / right ~40% of the video	Seek −10s / +10s with the accumulating indicator
Mobile	Double-tap middle ~20%	Nothing (dead zone, like YouTube)
Mobile	Single tap in the same zone while the indicator is visible (~0.9s)	Keeps seeking: 10s → 20s → 30s
Desktop (MaterialDesktopControls)	Double-click anywhere on the video	Toggle fullscreen (like the YouTube desktop player)

New ChewieController options, all default-on:

• allowDoubleTapSeek (default true) — mobile double-tap seek zones.
• doubleTapSeekDuration (default 10s) — the per-tap seek step.
• allowDoubleTapToggleFullScreen (default true) — desktop double-click fullscreen; only honored when allowFullScreen is true.

Double-tap seek is disabled on live streams and when playback is finished, and seeks are clamped to [0, duration] via the existing _seekRelative.

How

• The keyboard seek indicator state that feat: YouTube-style seek indicator for keyboard seeking #953 added inline to _MaterialDesktopControlsState moves into a shared SeekIndicatorStateMixin (lib/src/helpers/seek_indicator_mixin.dart), so both control sets drive the same SeekIndicator widget and share the accumulate/reset/auto-hide logic.
• MaterialControls._buildHitArea wraps its GestureDetector in a LayoutBuilder; onDoubleTapDown records the tap's dx and onDoubleTap dispatches by zone (< 0.4 back, > 0.6 forward, middle dead). The double-tap recognizers are only attached while double-tap seek is applicable, so single-tap latency is unchanged when the feature is off or the stream is live.
• Tap-to-repeat: the single-tap handler first checks whether the indicator is visible and the tap landed in the active zone; otherwise it falls through to the existing show/hide-controls / play-pause logic untouched.
• Desktop gets onDoubleTap → _onExpandCollapse on its hit area. Desktop single-click behavior is deliberately unchanged — pauseOnBackgroundTap already lets apps opt into click-to-pause.

Notes

• Double-tap seek is credited to @punit1111's Added seek forward and rewind buttons #717, which first explored it — this PR implements it on top of current master with clamping, live/finished gating, the dead zone, and the shared indicator.
• Known limitation: while the controls are hidden, the first tap reveals them and only then do double-taps seek (the hit area sits under the controls' AbsorbPointer). Matching YouTube's double-tap-while-hidden would need the recognizer above the absorber, which would add tap latency to every control button — out of scope here.

Tests

test/double_tap_seek_test.dart adds a fake VideoPlayerPlatform (modeled on video_player's own test fake, hence the video_player_platform_interface dev dependency) so the controller really initializes and positions can be asserted. Covers: both zones seek and clamp, the configured step, the dead zone, live/flag gating, tap-to-repeat (same zone accumulates, opposite zone falls through, faded indicator does not repeat), single-tap regression, desktop double-click fullscreen on/off, and option defaults + copyWith round-trips.

dart format, flutter analyze lib test, and the full flutter test suite (105 tests) are clean.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 79.47%. Comparing base (00d4b3a) to head (62d9d8f).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #964      +/-   ##
==========================================
+ Coverage   77.70%   79.47%   +1.76%     
==========================================
  Files          43       44       +1     
  Lines        2687     2738      +51     
==========================================
+ Hits         2088     2176      +88     
+ Misses        599      562      -37     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[diegotori]

@Ortes you still have changelog conflicts in this PR. Thanks.

[diegotori]

@Ortes looks like every time I merge your PRs, it always conflicts with the changelog entries that you make ahead of time. Please re-sync them. Thanks.

[Ortes]

@diegotori The conflicts are structural rather than stale branches: every PR inserts at line 1 of CHANGELOG.md, so any two of them conflict by definition, and the ## [Unreleased] → ## [1.16.x] rename at release conflicts on that line too. With 8 PRs open it re-triggers on every merge.

Since you re-file the entries under the release version anyway (#959 → 1.16.2, #966 → 1.16.0, #970 → 1.16.3), would you rather I drop the CHANGELOG hunk from my PRs entirely and put the proposed entry in the PR description instead? That removes the conflict class completely. Happy to keep writing them if you prefer — just say which.

Meanwhile I'll re-sync all the open ones now.

[diegotori]

@Ortes, usually I have Gemini create the changelog entries before publishing. Ideally, I usually take point on releasing new versions to pub.dev.

If you want, you can omit those messages from your PR so that I can update it when I publish it.

[Ortes]

@diegotori Unrelated to this PR, and easy to say no to: would you consider adding me as a collaborator on chewie? I asked in #953 and #946 but it probably got lost in the PR noise.

Where I'm at: 17 PRs merged, 7 of them in the last two days. This week I re-synced the whole open stack onto the 1.17.x cast refactor, and dropped the CHANGELOG entries from my PRs as you asked — so that particular friction is gone from your side.

What I'd use it for: triaging incoming issues and PRs (labels, duplicates, asking for reproductions) and reviewing other people's PRs, so the queue moves without everything waiting on you. Releases and pub.dev publishing stay yours — I'm not asking for those. Merge rights only if and when you'd want that.

If it needs the fluttercommunity org rather than you, tell me the process and I'll do the legwork. And if the answer is no, or not now, just say so and I'll drop it.

[diegotori]

You need to add some test coverage before we can move forward with this.

Thanks.

[diegotori]

@diegotori Unrelated to this PR, and easy to say no to: would you consider adding me as a collaborator on chewie? I asked in #953 and #946 but it probably got lost in the PR noise.

Where I'm at: 17 PRs merged, 7 of them in the last two days. This week I re-synced the whole open stack onto the 1.17.x cast refactor, and dropped the CHANGELOG entries from my PRs as you asked — so that particular friction is gone from your side.

What I'd use it for: triaging incoming issues and PRs (labels, duplicates, asking for reproductions) and reviewing other people's PRs, so the queue moves without everything waiting on you. Releases and pub.dev publishing stay yours — I'm not asking for those. Merge rights only if and when you'd want that.

If it needs the fluttercommunity org rather than you, tell me the process and I'll do the legwork. And if the answer is no, or not now, just say so and I'll drop it.

I'll have to think about this. You'll have to go through them, and we'll need to make sure not to step on each other's toes.

Once they speak to me about it, then we'll see what happens.

[Ortes]

@diegotori Added test/double_tap_seek_test.dart coverage for the branches codecov flagged as missing in material_controls.dart (the 8 uncovered lines in the patch): pauseOnBackgroundTap on the hit area, the display-tapped controls-hide toggle, and the backward direction of the tap-to-repeat zone check. flutter analyze and the full flutter test suite (197 tests) are clean.

[diegotori]

@Ortes looks like you still have conflicts to deal with. Thanks.

[diegotori]

LGTM.