Remove the parameter body_aero from the constructor of the Solver

[ufechner7]

Writing Solver(body_aero; ...) gives the impression that body_aero would be stored in the Solver struct, which is not the case. This is confusing. Instead, a Settings struct should be used to define the required dimensions.

[1-Bart-1]

Writing Solver(body_aero; ...) clearly shows that a body aero and solver are co-dependent. You cannot use one solver on different body_aeros with different amounts of panels. I don't find this confusing.

[1-Bart-1]

Adding a new constructor using a settings struct is ok, but please don't remove the Solver constructor with body_aero as an argument.

[1-Bart-1]

I propose having two Solver constructors:

Solver(settings)
Solver(n_panels, n_groups)

A check is needed when running solve, to make sure the n_panels and n_groups are correct.

[ufechner7]

I would suggest:

Solver(se::VSMSettings; kwargs)
Sover(n_panels, n_groups; kwargs)

[1-Bart-1]

I agree

[1-Bort-1]

Deprecating Solver(body_aero; kwargs...) and Solver(body_aero, settings): they still work but warn once per call site (Base.depwarn, force=true) and point to Solver(settings) / Solver(n_panels, n_unrefined_sections). Builds on #340. Moving make_dual_shadow, examples, docs and tests to the new constructors so none of them warn; one test checks the warning.