feat: [Remote rendering 3.2c] attribute camera changes to the input that caused them

[LKasianAnsys]

Issue

Addresses #20

Context

This is the third of 4 PRs for user story 3.2 (#20) of the phased implementation plan (ADR here) for adding remote rendering in VISOR.

After PRs 3.2a (#110) and 3.2b (#111) the server holds the camera and reads it on save_state, but only the load_state path writes it. For the record to stay current the browser needs to report camera changes back, and for the server to accept those reports the client needs to distinguish a user gesture from a camera change the server pushed; otherwise, every server-side camera update is echoed back as a gesture. This PR tags (on the client side only) each camera change event with its origin, user input or programmatic:

• CameraGestureTracker tags a camera event gesture only while a mouse button is held on canvas, or within 300ms of a wheel event or a z/r keydown, and settles 300ms after the last event. Anything else is programmatic. (see the CameraGestureTracker docstring )
• It settles 300ms after the last camera event and reports the origin once per settle.
• IRenderer gains addCameraSettledListener; both renderers implement it.
• VtkScene constructs the tracker and forwards camera events to it.

Nothing subscribes to the settle yet, so runtime behaviour is unchanged; the subscriber and the trigger to the server are implemented in 3.2d.

Testing

Manual tests of this PR need the wiring added in 3.2d, so the two need to be tested together.

---

Copilot summary

This pull request introduces a new mechanism for debouncing and attributing camera change events to either user gestures or programmatic actions. The main addition is the CameraGestureTracker, which tracks user input and reports when the camera has "settled" after a change, along with the origin of that change. This enables more accurate server-side camera state recording and improves the clarity of user-initiated vs. programmatic camera changes. The implementation includes new tests, interface changes, and integration into both the renderer and scene classes.

Key changes:

Camera gesture tracking and debounce logic

• Added CameraGestureTracker (wasm/CameraGestureTracker.js) to debounce raw camera events and attribute each settle window to either a user gesture or a programmatic change, with detailed handling of user input and event timing.
• Comprehensive Jest tests for CameraGestureTracker, verifying debounce timing, input attribution, and teardown behavior.

API and type changes

• Introduced a new CameraOrigin type ('gesture' | 'programmatic') and documented its semantics in IRenderer.ts.
• Added addCameraSettledListener to the IRenderer interface, providing a callback that fires once per settle window with the origin of the change.

Renderer and scene integration

• Implemented addCameraSettledListener as a no-op in NullRenderer and as a proxy to the underlying scene in WasmRenderer. [1] [2]
• Integrated CameraGestureTracker into VtkScene, ensuring proper listener management and teardown during scene rebuilds to avoid stale reports. [1] [2]

[ansBAkula]

Tested. Thanks.

The base branch was changed.