Support observable list for `expect`

[ytdHuang]

Problem Description

For the current implementation of expect, it only allows a single observable (Operator).
We should also support it to be the a list of observables.

Furthermore, I saw some methods of expect allow AbstractQuantumObject.
Is it reasonable to calculate expect for QobjEvo ?
I thought it will always return the value of zero. Because it uses the default value of ScalarOperator, which is 0.0 + 0.0im.

Maybe we should just support Qobj for expect

Proposed Solution

Just add more methods.

Alternate Solutions

No response

Additional Context

No response

[albertomercurio]

We should also support it to be the second argument.

Why should we support the observable to be the second argument? The qutip documentation seems to define it only as a first argument, which makes sense for me.

Is it reasonable to calculate expect for QobjEvo ?
I thought it will always return the value of zero.

If the QobjEvo called update_coefficients!, then this would not necessarily be zero. In that case the meaning is to compute the expectation value at the last computed time. Maybe we can write some comments about that in the docstring, but I think that this functionality may be useful sometimes.

[ytdHuang]

@albertomercurio
I think it's just that qutip also supports specifying a list of observables.

I will make a PR later.

Regarding the QobjEvo for expect, it's okay for me.

[ytdHuang]

The observable list might contain different data types of Qobj, and thus, the return value might be Float64 or ComplexF64 (type instability).

[albertomercurio]

But what if we have both operator list and state list?

I that case I think we return a matrix. Is that right? We can avoid type instabilities by returning a Float64 matrix only if all the operators are hermitian, and ComplexF64 otherwise.