Fix limits on angles and angular dispersity

[pkienzle]

• The θ angle ranges over [-90°, 90°] for both magnetic and nuclear angles. Limit the range to [-180°, 180°] in all cases. This allows for fits to converge smoothly even when the value is near a pole, though it does mean there are 2 (or 4 depending on symmetry) equivalent solutions in the space.
• Limit the range on φ for the magnetic moment and the applied field to [-360°, 360°] to match the limits for φ in the shape. Similar to θ, this is double the natural range of the parameter.
• Angular dispersity limits for the bumps wrapper were wrong. Dispersity should be non-negative.
• Update docs to explain why the default limits are twice what you might expect.

The up_theta parameter defaults to 90° as it should since an applied field along the beam direction is an unusual configuration.

The sld_mtheta parameter defaults to 0º. Is this reasonable? Or should it, too, default to 90º since sample magnetism is likely to be aligned with applied field?

Note: I still don't have a good fit to the magnetic example from the sasdata repository.

Warning The git commit hook did lint changes on the 2-Yukawa model even though it was not part of the PR.

[butlerpd]

As discussed at today's fortnightly call, this is likely to be important to the hard matter group at contributor camp. @dehoni has agreed to take a look.