usdAbc: do not collapse Alembic geometries into their transform parent if the parent has multiple children

[mattyjams]

Description of Change(s)

This avoids erroneously collapsing multiple sibling geometries into a single USD prim, forcing one to clobber the others.

The Alembic file used by the unit test contains six (poorly hand/mouse-drawn) curves representing numerical digits. The structure of the file looks like this (as reported by abcinfo -o):

testUsdAbcXformPrimCollapsing.abc, version: 10705 (2 timings)
   TimeSample[0]      1 samples:  t0=0 (frame 0.0 @ 24fps)
   TimeSample[1]      1 samples:  t0=0.0416667 (frame 1.0 @ 24fps)

Objects:
/CurvesAsset (Xform)
/CurvesAsset/Geom (Xform)
/CurvesAsset/Geom/Curve1 (Xform)
/CurvesAsset/Geom/Curve1/CurveShape1 (Curves)
/CurvesAsset/Geom/Curve2 (Xform)
/CurvesAsset/Geom/Curve2/CurveShape2 (Curves)
/CurvesAsset/Geom/Curve3 (Xform)
/CurvesAsset/Geom/Curve3/CurveShape3 (Curves)
/CurvesAsset/Geom/Curves456 (Xform)
/CurvesAsset/Geom/Curves456/CurveShape4 (Curves)
/CurvesAsset/Geom/Curves456/CurveShape5 (Curves)
/CurvesAsset/Geom/Curves456/CurveShape6 (Curves)

12 Objects displayed

Prior to this change, the resulting Curve1, Curve2, and Curve3 BasisCurves prims are all correctly translated and collapsed from their respective Alembic Xform and Curve nodes through the file format plugin, but the Curves456 BasisCurves prim would only represent the CurveShape6 curve from Alembic and would have lost CurveShape4 and CurveShape5 from having attempted to merge all three curve nodes into a single transform parent.

With this change, a parent with multiple children like this is detected as a case where collapsing cannot be performed, and Curves456 ends up as an Xform prim instead with three BasisCurves children.

• [ X ] I have submitted a signed Contributor License Agreement

[sunyab]

Filed as internal issue #USD-7480

[tallytalwar]

Hi @mattyjams

Just wanted to note that, while running usdAbc tests internally, the testUsdAbcInstancing* tests failed. But I believe these are more correct now and their baselines should be updated. I am going to include the updated baselines for testUsdAbcInstancing* tests with this change.

Thanks

[mattyjams]

Yikes! Clearly I didn't run the full battery of tests...

Sorry about that, but thanks for catching it, @tallytalwar!