Deferred evaluation of uncertainty values

[c3d]

When we have expressions for standard uncertainty, we currently evaluate all the way.

This comment suggests that it could be a good idea to defer evaluation.