1. it allows your camera script to exist without having a reference to a player. Which might be useful sometimes.
2. camera node does not have to be a child of the player node (although it can still be if you want)
3. for first person games, it allows you to attach a legs/feet mesh to your player without it being dependent on the rotation of the camera.
4. for third person games, it allows for more smooth camera movement where the camera isn't actually attached to the player (2). It also allows the player to have a different rotation from the camera which is very common for third person games.
5. it allows you to use the same camera script for first person games and third person games. (and other types of games probably)
It's all kind of small benefits that can be easily worked around. In the end it comes down to personal preference more than anything :3