Add support for dual_ftol_rel parameter in NLopt

[ChrisRackauckas]

Summary

Fixes #922 - This PR adds support for the dual_ftol_rel parameter in OptimizationNLopt, which is used by MMA/CCSA algorithms to control the dual optimization subproblem tolerance.

Problem

When trying to use the dual_ftol_rel parameter with NLopt solvers like LD_CCSAQ, users would get an error:

sol = solve(prob, NLopt.LD_CCSAQ(), xtol_rel = 1e-7, dual_ftol_rel = 1e-7)
# ERROR: UndefVarError: `dual_ftol_rel\!` not defined

This happened because the code was trying to call NLopt.dual_ftol_rel\!() which doesn't exist. The dual_ftol_rel is not a direct setter function but a named parameter in NLopt.

Solution

Added special handling for dual_ftol_rel in the __map_optimizer_args\! function:

• Check if the keyword argument is dual_ftol_rel
• If yes, use NLopt.nlopt_set_param(opt, "dual_ftol_rel", value) instead of the generic eval approach
• Otherwise, continue with the existing behavior

Why this matters

From the NLopt documentation:

Because this subsidiary solve requires no evaluations of the user's objective function, it is typically fast enough that it can be solved to high precision without worrying too much about the details. However, in high-dimensional problems you may notice that MMA/CCSA is taking a long time between optimization steps, in which case you may want to increase dual_ftol_rel or make other changes.

This parameter is particularly important for users working with high-dimensional optimization problems using MMA or CCSA algorithms.

Testing

Verified that:

• The parameter can now be set without errors
• The fix uses the correct NLopt API (nlopt_set_param)
• Backward compatibility is maintained for all other parameters

🤖 Generated with Claude Code

[ChrisRackauckas]

Needs a test.