Add speech recognition permission type

Before an application can access on-device speech recognition it needs to ask at the very least for permission to access the microphone, perhaps also the speech recognition service itself.

For cloud-hosted speech recognition, getting permission from the end-user is even more important as speech data is transmitted to 3rd parties over the network. This has implications for privacy, battery-life, and network data volume.

Qt applications using those services need to declare their access to such guarded services via the respective manifest file, and then also ask for permission from the user. This needs to be integrated into the Qt Permission API (QTBUG-90498) framework we are introducing in Qt 6.5.