perf(Renderer): accumulate actors and volumes without concat

[PaulHax]

Context

getActors() and getVolumes() collect each prop's objects using repeated
concat calls. Each concat call copies the growing result array.
This quadratic work repeats on every getActors call, including getActors()
during picking without an explicit pick list.

Batching geometry into fewer actors is the vtk.js way, but separate
actors are useful when datasets load and unload independently. This change came
out of loading tiled point clouds, where chunks of points are swapped in and out as the view changes.

Results

Recorded headless Chromium medians per collection call, spanning actor and
volume results:

Count	Before	After
400	0.05–0.06 ms	0.005–0.010 ms
2,000	1.08–1.21 ms	0.035–0.055 ms

At 2,000 objects, this saves roughly 1 ms per call.

Changes

• Use a shared push helper for linear collection, preserving array and
  non-array prop answers.
• Preserve order and fresh snapshots; by-reference getters return the latest
  snapshot.

PR Checklist

• GitHub Actions CI passed: semantic-release commit messages, lint, and tests
• Test coverage added
• Documentation and TypeScript definitions are updated to match these changes