Camera apps on Android are very loosely coupled to the OS. They are intentionally left to OEMs to provide because that's the most visible aspect of hardware differentiation, and that differentiation probably depends on software support. On top of that, it would be hard to design an API for every possible camera hardware, apart from a high level API for apps to acquire an image.
>There are only a few ways that cameras themselves work, and even few ways that cell phone cameras work.
Have an infrared camera that augments the image from a normal camera?
Have a rotationally pop up camera that allows using the same 3 cameras for back and selfie, but also use it to take panoramas. (I miss my Asus ZenPhone flip)
Create photos that allow users to change focus when viewing?
Have two cameras back to back and allow capturing simultaneously to create 360 photos/videos?
Have two cameras side by side and allow stereo vision?
If you have a 0.7x, 1x, and 4x camera, and the user is zooming at 3, use the 1x to fill the frame, but the 5x to have better image quality at the center?
Use the optical stabilizer to take several shots with micro shifts and do super resolution?
We can go on and on. Cameras and even smartphone cameras allow a lot of possibilities. Some of them are already explored by some manufacturers.
>On top of that, it would be hard to design an API for every possible camera hardware, apart from a high level API for apps to acquire an image.
APIs themselves are hard to make, but why is a camera one especially so? The language is well understood, the math and science are well understood. There are only a few ways that cameras themselves work, and even few ways that cell phone cameras work.
Why is it hard?
In advance -- No, Sony/Panasonic/Toshiba/Apple/Whoever locking functions behind magic numbers and proprietary blobs and other 'un-Gentlemanly' things shouldn't count as difficulty in making a Camera API; that's just shit companies being shit to people, not an API problem.