Diff of a ParameterFunction doesn't work when evaluated at a constant

Describe the bug When defining a parameter as the diff of a ParameterFunction sometimes it returns 0. I think the issue is that the expression gets simplified first and, because the argument is a constant, we loose track of the differentiation variable. Then, when taking the derivative, the variable is not in the tree and therefore it returns 0.

To Reproduce Steps to reproduce the behaviour:

1. Add a diff parameter to your model, for example

            def Uprime(sto, T):
                inputs = {"Positive particle stoichiometry": sto}
                Uprime_ref = pybamm.FunctionParameter(
                    "Positive electrode OCP [V]", inputs, diff_variable=sto
                )
                return Uprime_ref  / pot

1. Set the positive OCV from a function (for example a linear function).
2. Evaluate at c_0 = param.c_p_init(1), the function returns zero.
3. Evaluate at c_0 + 0 * pybamm.t, the function now returns the right value.

[brosaplanella]

Ok, then I will implement the hacky method of the Multiplication by one, as you did with c_e.

[tinosulzer]

You might need to do the adding 0 * pybamm.t for it to work. Otherwise even the multiplication will get simplified out later, e.g. 2 * pybamm.Multiplication(5, 1) would realize that the whole thing is constant and return Scalar(10)