Unify steering studies under Maneuver

[rhorvath02]

What

Replaces the separate steering entries in the simulation catalog with Maneuver. Setup selects DynPy 3/6/10/14 DOF or BobLib MBD, then ramp and hold, step, or sine steering. A live waveform preview updates with magnitude, start time, duration or frequency. Only the selected waveform's controls are shown.

DynPy inputs use nominal roadwheel degrees. BobLib inputs use handwheel degrees, with labels and the explanation showing that distinction. Existing saved studies and CLI entry points remain available. New runs retain reports, signals and result post-processing.

How to check

• Browser checks passed for the catalog, live sine preview, conditional fields and switching between MBD and DynPy while preserving waveform settings.
• Actual DynPy runs passed at 3, 6, 10 and 14 DOF. Recorded steering commands matched the input waveforms.
• Actual MBD ramp, step and sine runs passed in Docker and produced reports.
• The 139 targeted app, archive and waveform tests passed. Updated catalog and editor checks pass.
• Full local CI passed with 1062 tests passed and 21 skipped. The 15 waveform tests also passed after the signed-amplitude update.
• The isolated Modelica baseline completed with 18 passed, one skipped and the same two four-post/ramp-steer failures as the preceding audit. All ten reported mismatches are identical. No baseline was changed.
• The final MBD sine report was rerun successfully with the existing startup-cycle exclusion preserved.
• GitHub container CI cannot fetch the OpenModelica signing key because build.openmodelica.org times out. The cached local Docker toolchain completed the maneuver checks and baseline rerun.

Notes

Stacked on #140. A sine maneuver evaluates one frequency. It is not an automatic frequency sweep. Ramp offset is represented by its start time.