disables options menu to avoid leak
 removes on click listener to avoid leak

The test class left SearchStateKt mocked and caused other tests to fail intermittently when
running after it.

When the user opts out of telemetry, we need to reset the user identifiers
used by the experiment system, so that there's no risk of tracking the
user across the reset event if they later decide to re-enable telemetry.

Speculative fix based on the Firebase logs which shows that on a second check
of R.id.tab_wrapper it is not found the screen.
It may be because after `advanceToHalfExpandedState` and before getting a
reference to it in `waitForTabTrayBehaviorToIdle` it was already animated off
the screen.
With this in mind I've added a null check for the view reference before trying
to register an idling resource on it's Behavior.

Also added and used a way to click at a specific location in a View, not just
in the default middle in the View.
It was observed from the Firebase videos that a "click" on the topBar actually
selected the private tabs section. This would leave us to believe that the
"click" was caught by that other View which was placed above the x,y middle of
the topBar.

* For #17352 - Fixes deleted downloads reappearing

The `getDeleteDownloadItemsOperation` job would check fragment context not null after the fragment was stopped. Removing that check and only passing the downloadUseCase as a parameter fixes the problem.

Co-authored-by: Alex Ionescu <aionescu@mozilla.com>

Previously when the toolbar was on top the banner was inflated in the toolbar's parent - an AppBarLayout.
After migrating to use a custom behavior for scrolling the toolbar and not use
anymore the AppbarLayout for this we needed a new solution.
Using a new behavior to keep this banner in sync with the y translation of the
toolbar gives us most of the old behavior back.

* Feature flag for toolbar menu redesign. Add new items to menu and reorder.

* Handle toolbar items in menu controller

* Menu controller tests

* Make icons invisible

* Lint

* UI tests reflect design change

* Refactor test names

* Lint fixes

* UI tests

I think this is more consistent and centralizes the violation ignores.