Support propagation of uncertainties via uncertainties.ufloat throughout WEAC #47

We want to be able to propagate uncertainties through WEAC. I propose to accept uncertainties.ufloat inputs and make sure the solver works with ufloats.

Challenges

• I expect elastic properties, strength, and fracture toughness to be straightforward
• Segment lengths, crack lengths, and weak-layer thickness probably okay, too
• Slab layer thicknesses likely tricky