While improving Android IME support (da38c1ed), text editor destructors
were updated to explicitly pass keyboard focus elsewhere.
As far as I remember, the change was intended to prevent the text input
system from trying to send input events to components while they were
being destroyed, in which case the TextInputTarget and Component bases
may be 'valid', but the data members referenced by the TextInputTarget
implementation may have been destroyed.

The motivation for removing these lines is that giving away focus and
sending a focus event can cause all components to become unfocused. This
is problematic in the case of slider text editors - pressing 'enter'
will cause the TextEditor to be destroyed, but the parent component will
fail to gain focus, so pressing 'tab' will not have any effect.

Running under Gnome, calling ComponentPeer::forceSetBounds immediately
after creating a new peer causes the peer to emit a visibility-changed
event because the window manager sometimes reports that the new window
is in IconicState.

_NET_WM_STATE_HIDDEN seems to more accurately reflect whether the window
is really minimised or off-screen.

This makes the ordering consistent with other view modes of the
FileBrowserComponent and restores the behaviour prior to a400d3ebe0.

Previously, the top-level modal dismiss broadcaster would attempt to
dismiss all modals when the native file chooser was brought to the front
or clicked. This would end up calling Component::inputAttemptWhenModal,
which would bring the FileChooser's Component to the front, interrupting
the mouse click.

The fix is to avoid bringing the plugin window to the front when the
FileChooser is in a modal state. As the chooser is a native window, the
system should take care of enforcing the expected modal behaviour.

It's not clear why this issue only affects Cantabile. It seems that in
Cantabile, events from the native FileChooser get sent to the plugin
editor's parent HWND, whereas this doesn't happen in other hosts.
Perhaps Cantabile is hooking/intercepting window messages in some way.

This fixes a regression introduced in
a400d3ebe0.

One issue affected the situation where the provided bounds wouldn't
start at (0, 0). Such bounds are regularly acquired by calling
Rectangle::reduced().

The other issue affected the width calculation of fractional items.
The error wasn't correctly integrated during the computation, and as a
consequence the last fractional element would exhibit all the
accumulated error.

Prior to this commit all Grid calculations were carried out using
floating point numbers. The dimensions of all items would then be
rounded with the same function to calculate the integer dimensions used
for Component layout. This resulted in layout solutions where the width
or height of items with dimensions specified using the absolute Px
quantity could differ from the correctly rounded value of these values.

This commit ensures that the width and height of these items are always
correct and their cumulative error in size is distributed among items
with fractional dimensions.

This commit fixes a bug where a View would be registered as it's own
parent in the accessibility hierarchy leading to stack overflows in the
Android accessibility service. The issue affected up-to-date Android 13
systems and potentially older systems with updated accessibility
services.

Previously when using the FileBrowserComponent in TreeView mode, a
refresh would delete all items and rebuild the UI based on new
directory scan data, losing the openness state in the process. With
this commit only changes are applied to the current TreeView.

When clearing deferredRepaints in drawRect, dirty regions in the 'next'
frame could be incorrectly removed while drawing the 'current' frame.
Now, deferredRepaints is cleared in setNeedsDisplayRectangles, as it
was before 9d50ab6c59.