Uncertainties from beam center and direct beam

Now that we are computing both beam center and direct beam fully with Scipp, and that we use the upper-bound method when broadcasting, we should put the uncertainties back into the operations.

There is still some uncertainty on how to proceed when we resample the direct beam to a different wavelength resolution.