standalone apps

This change fixes a crash in Reaper when closing a plugin window
with ALT + F4.

Fixes an issue introduced in 72fa2d98
where zeromem might write past the end of the channel buffer.

NSViewComponentPeer:  Avoid making temporary windows (dropdowns, callouts) full-screen when in full-screen mode

eabcfbad incorrectly set the collection
behaviour NSWindowCollectionBehaviorFullScreenPrimary on all top-level
components, which would cause popup menus and callouts to appear in
their own space when launched from an app running in full-screen mode.

This fixes an issue where REAPER was unable to completely unload VST3
plugins.

The host is guaranteed to re-scan all parameter values after setting a
new state on the plugin, so there's no need to notify the host about
parameter changes that happen during the setState call.

This also avoids a problem where Bitwig would complain about invalid
parameter IDs in 'engine.log' when loading projects containing JUCE
VST3s. These log messages were produced because a call to setState was
made before Bitwig registered the plugin's parameters, and the setState
call in turn called performEdit with parameter IDs that were yet to be
registered.

If the block size changes from block to block, then it's possible for
inputBuffer to be smaller than buffer, but for buffer to be smaller than
the initially-allocated size of inputBuffer.

The previous implementation would pass the mouse wheel event up to the
component's parent, as long as the parent was enabled. This meant that a
wheel event on the innermost component of a hierarchy such as
"[[disabled] enabled]" would send the event to the parent, but a wheel
event on the innermost component of a hierarchy such as
"[[[disabled] disabled] enabled]" would 'eat' the event and prevent it
from propagating.

After this change, unhandled mouse wheel events will always be passed to
the nearest enabled parent. This behaviour is more consistent and
intuitive.

PluginDescription:  Fall back to using deprecatedUID in createIdentifierString() if the new uniqueID is not known

This button is used to access hidden toolbar items, as well as hidden
tabs. The old "tabs" name was misleading when the button was located in
a toolbar.

Previously, setting a new look and feel would create a new missing-items
button, but would not add it to the toolbar or set the onClick callback.

CFAttributedString ranges must be given in terms of 16-bit word offsets,
rather than codepoints.

In CoreGraphicsPixelData::createImage, image data was copied from a
BitmapData created from the Image passed into the function.

The BitmapData instance didn't keep track of the size of the buffer it
pointed to, so the buffer size was computed by multiplying the
BitmapData height by its line stride. However, if the BitmapData pointed
to a subsection of an image, the `data` pointer might be offset from
the allocated region, and `data + lineStride * height` would point past
the end of the allocated region. Trying to read/copy this range would
cause a heap buffer overflow at the end of the range.

This change adjusts BitmapData so that it keeps track of the size of the
allocated region. Taking a subsection of an image should subtract the
data pointer offset from the size of the allocated region.

This mirrors the behaviour on macOS.

In OpenGL 3 and up, GL_EXTENSIONS is deprecated as an argument of
glGetString and glGetStringi should be used instead.

Showing the tip will in turn call getDesktopScaleFactor(), accessing the
lastComponentUnderMouse. In some cases, it was possible for
lastComponentUnderMouse to point to a deleted component, resulting in
UB.

There are two changes in this PR:
- Using a SafePointer rather than a raw pointer ensures that calls to
  getDesktopScaleFactor() will always be safe, regardless of when they
  happen.
- Moving the assignment of lastComponentUnderMouse to before the call to
  displayTipInternal() ensures that the returned scale factor is that of
  the component that the mouse is currently hovering.

VST3 Host:  Avoid calling initialize twice on objects that implement both IComponent and IEditController

Narrator on Windows seems to call ExpandToEnclosingUnit to find the
character immediately following the current cursor position. When
expanding a degenerate range with position `input` (used to represent
the cursor position) to a single-character range, the returned range
should start at `input` and end at `input+1`. Previously,
findTextBoundary() would always return the position before
currentPosition when searching backwards by character, so the result of
ExpandToEnclosingUnit would be off-by-one when expanding to the closest
character.

Previously, the child process could be deleted from its own
handleConnectionLost() callback, which would attempt to stop the child
process's thread from that same thread.

Now, the demo will attempt to stop the coordinator thread from the main
thread instead.