Blocking: the Phase 548 hit-test contract currently treats freehand primitives as filled rectangles.

GraphWhiteboardPrimitiveRendererAdapter.HitTest(...) sends every primitive through Contains(...), and Contains(...) only checks BoundsOrigin + BoundsSize. The freehand points projected into GraphWhiteboardPrimitiveSceneItem.Points are never used, so any click inside the freehand bounding box returns a hit even when it is far from the actual stroke. That would give future selection/edit/eraser paths false positives while the docs/tests claim topmost primitive hit-testing evidence for rectangle/freehand primitives.

Please either make HitTest kind-aware for freehand primitives, for example using segment-distance plus an explicit tolerance, and add a negative test for a point inside the freehand bounds but away from the stroke; or narrow the Phase 548 claim/tests so this slice does not present freehand hit-testing as proven.

Addressed the freehand hit-test blocker in 3dad5ae.

What changed:

• Added regression coverage for a point inside the freehand bounds but away from the stroke returning no hit.
• Added positive coverage for a point near a freehand stroke segment returning the primitive hit.
• Updated GraphWhiteboardPrimitiveRendererAdapter to keep rectangle hit-testing bounds-based while freehand hit-testing uses bounds as a fast reject and then segment-distance tolerance against the projected points.

Verification:

• dotnet test tests\AsterGraph.Editor.Tests\AsterGraph.Editor.Tests.csproj -c Release --no-restore --filter FullyQualifiedName~WhiteboardPrimitiveRendererAdapterContractsTests --verbosity minimal => 5/5 passed
• dotnet build AsterGraph.sln -c Release -v:minimal => 0 warnings, 0 errors
• git diff --check => exit 0, CRLF notices only
• GitHub checks on PR Phase 548: whiteboard primitive renderer adapter skeleton #220 head 3dad5aecd9ed7f4191a0a20073a8f2bdd6d347b1 are all passing, including hygiene-proof, framework matrix, linux/macos validation, contract-proof, and release-validation.