skip solved2dof benchmarks when pyOptSparse is unavailable

[vyom-aggarwal]

Both tests in test_bench_solved2dof.py drive the problem with IPOPT, which requires pyOptSparse, but the file never imported or applied @require_pyoptsparse. Without pyOptSparse installed they raise "ImportError: pyOptSparseDriver is not available" and fail, rather than skipping the way every other pyOptSparse-dependent benchmark does.

This is the only benchmark file that uses a pyOptSparse optimizer without the decorator; the 20 others already guard their IPOPT/SNOPT tests this way.

Summary

aviary/validation_cases/benchmark_tests/test_bench_solved2dof.py runs both of its benchmarks with the IPOPT optimizer (optimizer='IPOPT' in test_bench_Solved2DOF, add_driver('IPOPT', ...) in test_bench_Solved2DOF_landing), which requires pyOptSparse. The file never imported or applied @require_pyoptsparse, so in an environment without pyOptSparse both tests fail:

ImportError: pyOptSparseDriver is not available, pyOptsparse is not installed.
This is the only benchmark file that uses a pyOptSparse optimizer without the decorator. The other 20 benchmark files already guard their IPOPT/SNOPT tests this way, and test_subsystems_within_a_mission.py correctly needs no guard because it uses SLSQP. This was not a regression: IPOPT has been used here since the file was added in 529de2f, and the decorator was never present.

This PR adds @require_pyoptsparse(optimizer='IPOPT') to both tests so they skip cleanly, matching the existing convention in test_bench_GwGm.py.

Effect on a full testflo . run without pyOptSparse: 2 failed / 46 skipped becomes 0 failed / 48 skipped. On a machine that does have pyOptSparse with IPOPT, behavior is unchanged: require_pyoptsparse returns the test object untouched when the import and OPT(optimizer) both succeed, so these benchmarks still run and still assert as before.

Related Issues

Resolves #
Backwards incompatibilities
None. No source behavior changes; this only affects test collection when an optional dependency is missing.

AI Usage

This change was assisted with Claude (Opus 5) via Claude Code. The AI model didn't code anything, but instead helped me get an idea of what I need to do.