Question What is Root Transform Rotation?

I'm reading the chapter on Root Motion in the Unity Docs and try to fully understand the concept of root motion. Whereas I imagine the Root Transform Positions (Y and and XZ) as a time series of vector positions in space, I'm horribly confused with the concept of Root Transform Rotation. That is, the actually confusing part is the description of the Based Upon property:

The first thing that confuses me is "orientation of the clip". What is meant by that? A clip is very simply put just a time series of rotations at the bone joints. How can that have a spacial orientation? The text further says that using Body Orientation causes motions like strafing to fail. Now, strafing is basically moving sideways. What does this have to do with rotation? To me moving sideways is more about Root Transform Position (xz).

In the editor, the options "Body Orientation" and "Original" are made mutually exclusive. But the description in the docs lets one think that they can be combined. To put it somewhat silly: Why can I not have the clip orientation to follow the forward vector of the body AND add the authored offset found in the imported clip?

There is a lot of confusion on my part. I hope someone can explain it in a better way.