DogAteMy, the easiest place to start is the moment of inertia tensor. We typically teach moment of inertia about a single axis, but what data do you need to compute it about an arbitrary axis?
No, that's contravariant, unless by gradient you mean the exterior derivative or differential.
Those are the coefficients of $df$. In Euclidean space with the usual boring inner product, it happens that they're the coefficients of the gradient, but that's an accident. So you really have to specify ...
I will just put it in my brain as existence of empty set is philosophically the same as just looking at an empty grass field. You know nothing exist in the grass field, but you can build things on it.