Call processed variables with custom spatial coordinates

[rtimms]

After #3234 ProcessedVariable objects now automatically take the names of the domains to plot automatically from spatial variables, but __call__ still expects the spatial variables to be x, y, z, r, or R, see:

    def __call__(self, t=None, x=None, r=None, y=None, z=None, R=None, warn=True):
        """
        Evaluate the variable at arbitrary *dimensional* t (and x, r, y, z and/or R),
        using interpolation
        """
        kwargs = {"t": t, "x": x, "r": r, "y": y, "z": z, "R": R}
        # Remove any None arguments
        kwargs = {key: value for key, value in kwargs.items() if value is not None}
        # Use xarray interpolation, return numpy array
        return self._xr_data_array.interp(**kwargs).values

We should make it possible to pass extra kwargs to __call__.

As a workaround, you can do something like

# variable that depends on a  spatial variable named "s" and time "t"
my_var_data_array = sim.solution["My var"]._xr_data_array

def my_var_eval(s, t):
    return my_var_data_array.interp(s=s, t=t).values

[brosaplanella]

Yes! I thought of this when doing #3234 but didn't open an issue for it.

[Saswatsusmoy]

Can we do something like this?

-Allow passing arbitrary keyword arguments to call by adding **kwargs
-Get list of expected args from the method signature
-Get args passed to call and kwargs
-Construct coord dict from expected args that were passed
-Update with any extra kwargs
-Pass coord dict to interp

[rtimms]

@Saswatsusmoy sounds good, would you like to give it a go?

[Saswatsusmoy]

@rtimms I have tried and modified the call method to accept **kwargs as a parameter. This will allow any additional keyword arguments to be passed to the method.

[rtimms]

@valentinsulzer id this something that can be fixed as part of #4600 ?