Flattop duration fix

[daveweisberg]

Fixed bugs in flattop_duration matching, namely to remove the dependence on coil_j_margin.

Previously, the output flattop_duration was forced be equal to the requirements.flattop_duration * (1 + requirements.coil_j_margin)

Now the output flattop_duration is exactly equal to requirements.flattop_duration.

• Calculation of flattop_flux_estimates should not be scaled by 1 + coil_j_margin
• c_flt should have a margin of 0.0, not coil_j_margin
• the margin for flattop_duration should not be dependent on coil_j_margin.

[orso82]

@daveweisberg regression tests are failing.

Also, I am not sure I agree with the change. If one wants to have say 20% tolerance on the current limit on the ohmic solenoid, I believe that's equivalent to say that they want to have 20% tolerance on their flattop duration. In other words, the nominal flattop duration is the one cited at 20% below the absolute maximum, no?

[daveweisberg]

@orso82 Ok, re-running tests with fix.

This change is important, because the previous way of doing this didn't match the main use case. What I want is a FUSE solution that has the requested flattop_duration when the CS coil current density is limited to the coil_j_margin. In practice, the CS coil would never be operated higher than this limit. So the actual flattop is never going to be longer.