Skeleton operator for the handoff / assembly scenes (COCO-17, not just a point)

[bamdadd]

The handoff_ltr and assembly_line scenes model the moving subject as a single point ("center"). Upgrade the moving subject to a COCO-17 skeleton (as in examples/assembly_station.py), so the scenes exercise pose across the overlapping cameras too.

Scope

• Attach a COCO-17 skeleton entity walking the same path instead of (or alongside) the point target, using the existing pose primitives (src/multicam_sim/pose.py, and the skeletal-motion work in Skeletal motion DSL: walk / reach / wave gaits for the COCO-17 skeleton #29 if landed).
• Keep a point-target variant or the skeleton's torso/center point so the existing coverage tests still have a single tracked point.

Acceptance

• The scene builds with a skeleton whose joints project into the cameras; coverage story (left loses it, right picks up, wide always sees) still holds for the torso/center.
• Deterministic, typed, ruff + mypy clean, tests updated.

Pointers

• examples/assembly_station.py (skeleton operator), src/multicam_sim/pose.py, issue Skeletal motion DSL: walk / reach / wave gaits for the COCO-17 skeleton #29 (skeletal motion DSL), src/multicam_sim/handoff_ltr.py.