The new camera implementation always bound all four CameraX use cases
(preview, image capture, video capture, and image analysis) regardless
of device capabilities. On devices with LEGACY camera hardware level,
this causes image capture to fail with "Capture request failed with
reason ERROR" because the hardware cannot handle that many simultaneous
use cases.
This change makes video capture and QR scanning use case binding
conditional based on CameraXModePolicy, which already determines device
capabilities. Video capture is only bound when the device supports mixed
mode (image + video simultaneously). QR scanning analysis is only bound
when explicitly requested.
Previously, the new camera would silently record video without audio
when microphone permission was missing. Now it shows the same rationale
dialog and permanent denial flow as the old camera.
