Bug with radius_effective_mode

[smk78]

Describe the bug
Looking into #3101 in more detail there appears to be a bug with radius_effective_mode which has been introduced since release 5.0.6.

Whilst the steps to reproduce might seem very niche, they are a set of steps that reliably reproduce the bug. There may be other sequences of steps that cause the same thing.

To Reproduce
Steps to reproduce the behavior:

1. Run SasView
2. Load some data (eg, 2% SDS); send to fitting
3. Create an ellipsoid @ hayter_msa model
4. Set radius_equitorial = 40
5. Tick background, radius_equitorial, volume_fraction & charge
6. Fit
7. Set radius_effective_mode = unconstrained
8. Fit
9. Set radius_effective_mode back to something other than unconstrained
10. Fit.

The radius_effective drops to zero!

And nothing you can do will recover it to a computed value based on the setting on radius_effective_mode, and the S(Q) remains frozen in subsequent fits.

Expected behavior
radius_effective should always take the value determined by the setting of radius_effective_mode.

SasView version (please complete the following information):

• 6.0.0b2

Operating system (please complete the following information):

• W10/x64

[krzywon]

This might be fixed with #3100. I tested your process in that branch and release_6.0.0. I can reproduce the issue in the release branch, but the issue seems to be solved in the PR branch.

[smk78]

Retested using https://github.com/SasView/sasview/actions/runs/10638002500 on W10/x64.
The issue described above appears to be fixed and the expected radius_effective is now updated to match the setting of radius_effective_mode.