Document in cirq.apply_unitary's docstring that it can operate on qudit subspaces

[Strilanc]

If you have a state tensor defined over qutrits, and you call cirq.apply_unitary targeting this state tensor but give it a qubit gate, apply_unitary will operate on the 01 subspace. This isn't currently documented but it should be mentioned in the docstring.

[aditya-giri]

@Strilanc I'd like to pick this up if that's okay - sounds like an easy first issue to get started with.

[aditya-giri]

Just to clarify with an example, since I'm new here - this means that if you have a single qutrit system (say, with states |0>, |1>, |2>, and currently in the superposition a|0> + b|1> + c|2>) - and you call apply_unitary but with a qubit gate instead of a qutrit gate, the unitary will only act on the a|0> and b|1> components and not on the c|2>. Is my understanding correct?

[vtomole]

@aditya-giri That's right.

[aditya-giri]

@vtomole does this also generalize to k-qubit gates applied to n-qubit systems (k<n)?