[macOS / iOS] Fix text-to-speech build with older macOS / iOS SDK.

Add __has_include check for AVFAudio include.
Add some explicit casts to avoid conflicts.
Change all `include`s to `import`s for consistency.

drivers/coremidi/midi_driver_coremidi.cpp

@@ -34,8 +34,8 @@
 
 #include "core/string/print_string.h"
 
-#include <CoreAudio/HostTime.h>
-#include <CoreServices/CoreServices.h>
+#import <CoreAudio/HostTime.h>
+#import <CoreServices/CoreServices.h>
 
 void MIDIDriverCoreMidi::read(const MIDIPacketList *packet_list, void *read_proc_ref_con, void *src_conn_ref_con) {
 	MIDIPacket *packet = const_cast<MIDIPacket *>(packet_list->packet);

drivers/coremidi/midi_driver_coremidi.h

@@ -36,7 +36,7 @@
 #include "core/os/midi_driver.h"
 #include "core/templates/vector.h"
 
-#include <CoreMIDI/CoreMIDI.h>
+#import <CoreMIDI/CoreMIDI.h>
 #include <stdio.h>
 
 class MIDIDriverCoreMidi : public MIDIDriver {