MindtPy follow-up: evaluate contrib.solver migration for direct continuous solves

[bernalde]

Follow-up from #3907 review comments: MindtPy currently routes continuous short-circuit QP/QCP solves through the existing APPSI and legacy solver capability logic. Reviewers noted that MindtPy should eventually move away from relying on APPSI-specific behavior and toward the newer contrib.solver interfaces, including the HiGHS quadratic support now available there.

This is intentionally separate from #3907 so that PR can remain focused on preserving current direct-solve behavior for models without discrete variables.

[bernalde]

Related follow-up from the review discussion in #3942: GDPopt result handling should probably be considered alongside the MindtPy contrib.solver migration work tracked here.

The immediate #3942 fix intentionally stayed narrow: LBB was calling the stale GDPopt private results helper on its time-limit path. Longer term, migrating GDPopt and MindtPy together would help keep result-object and termination-condition semantics aligned across the two decomposition solver stacks.