This adds a smoke test for every single API example. It also fixes 17 tests that had bugs in them, or were otherwise broken, and even fixes one actual bug in the framework, and one limitation in the framework.
The bug in the framework is that NetworkImage's _loadAsync method had await response.drain<List<int>>();, but if the response is null, it will throw a cryptic exception saying that Null can't be assigned to List<int>. The fix was just to use await response.drain<void>(); instead.
The limitation is that RelativePositionedTransition takes an Animation<Rect> rect parameter, and if you want to use a RectTween with it, the value emitted there is Rect?, and one of the examples was just casting from Animation<Rect> to Animation<Rect?>, which is invalid, so I modified RelativePositionedTransition to take a Rect? and just use Rect.zero if the rect is null.
This extracts the sample code out from the API doc comments, and places them in separate files on disk, allowing running of the examples locally, testing them, and building of slightly larger examples.
Switch document generation to use the snippets package instead of the snippets code in the Flutter repo. In the process, some bugs in sample code analysis have been fixed, and I've fixed some more errors in the samples.
This will allow the snippets package to be developed separately from the Flutter repo, and reduce the code in the Flutter repo.
The snippets code is deleted in this PR.
I also converted some comments in the snippet templates to be regular comments instead of doc comments, because having a doc comment block before the imports causes the Dart import sorter to lose the comment. They should have been regular comments in the first place.
The snippets package resides in the assets-for-api-docs repo.
The sample analysis has also been converted to be run in parallel, and I've bumped the Dartdoc version to 1.0.2.
The (new, not yet used) code gen for iOS was setting up a std::map from key codes to logical and physical key codes, but it was using uint32_t, which isn't big enough to hold the Flutter key codes.
Also, iOS needs to be able to filter out function keys, so I added a function key set.
This turns on order shuffling for all tests that don't fail with it on, marking those tests that do fail with a tag so that they will be run without shuffling on.
To determine which tests fail with it on, I ran all the tests 100 times with different random shuffle seeds, and then also ran it with the date seeds from today until the end of July, and tagged all of the test suites (files) that fail, with a seed that caused them to fail.
This adds avoid_dynamic_calls to the list of lints, and fixes all instances where it was violated.
Importantly, this lint is NOT turned on for flutter/packages/test, because those changes are happening in another PR: #84478
This is bassically reapplying #71721, but only enables it on web tests.
There are known issues that several tests under the `integration.shard`
depend on a specific platform, and as a result they require some
additional flexiblity (bots need to build more than one engine, and the
test flags should allow for secondary engines to be picked by such
tests).
By enabling this on the web-test shard, we will reduce the false
positives in the dart-flutter-HHH-web bot.
Fixing the more general problem is tracked by #72368.
