feat(mcap): add copy mode for multi-camera exports

Add an opt-in multi-camera copy layout for zed_svo_to_mcap.

copy mode preserves each camera's original timestamps and cadence,
writes namespaced /zedN topics without /bundle, and supports
common-overlap or full-range export windows.

Update the batch wrapper, Python validator, RGBD viewer, and
documentation so copy-layout MCAP files are treated as a first-class
format rather than invalid bundled output.

Validation:
- cmake --build build --target zed_svo_to_mcap -j4
- uv run python -m py_compile scripts/zed_batch_svo_to_mcap.py scripts/mcap_bundle_validator.py scripts/mcap_rgbd_viewer.py
- build/bin/zed_svo_to_mcap --segment-dir /workspaces/data/kindergarten/bar/2026-03-18T11-59-41 --output /tmp/bar_11-59-41_copy_common.mcap --encoder-device nvidia --depth-mode neural_plus --depth-size optimal --bundle-policy copy --copy-range common
- uv run python scripts/mcap_bundle_validator.py /tmp/bar_11-59-41_copy_common.mcap
- uv run --extra viewer python scripts/mcap_rgbd_viewer.py /tmp/bar_11-59-41_copy_common.mcap --summary-only

README.md

@@ -265,12 +265,51 @@ uv run python scripts/zed_batch_svo_to_mcap.py \
 
 The batch MCAP wrapper writes `<segment>/<segment>.mcap` by default, skips existing outputs unless told otherwise, and returns a nonzero exit code if any segment fails.
 The repo includes a minimal pose config at `config/zed_pose_config.toml` so MCAP conversion does not depend on a separate `cv-mmap` checkout.
-In bundled multi-camera mode, `--start-frame` and `--end-frame` mean the first and last emitted bundle indices from the common start timestamp, inclusive.
+In bundled multi-camera timeline mode, `--start-frame` and `--end-frame` mean the first and last emitted bundle indices from the common start timestamp, inclusive.
 When stderr is attached to a TTY, `zed_batch_svo_to_mcap.py` uses a `progress-table` view by default; otherwise it emits line-oriented start/completion/failure logs plus periodic heartbeat summaries. Use `--progress-ui table` or `--progress-ui text` to override the automatic mode selection.
 
-Bundled MCAP export now defaults to `--bundle-policy nearest`. That mode emits one `/bundle` manifest message per bundle timestamp on the common timeline and keeps the original per-camera timestamps on `/zedN/video`, `/zedN/depth`, and optional `/zedN/pose`. Consumers that care about grouping should follow `/bundle` instead of inferring bundle membership from identical message timestamps. Use `--bundle-policy strict` when you want the older thresholded sync behavior, and `--sync-tolerance-ms` only applies in that strict mode.
+Bundled MCAP export now defaults to `--bundle-policy nearest`. That mode emits one `/bundle` manifest message per bundle timestamp on the common timeline and keeps the original per-camera timestamps on `/zedN/video`, `/zedN/depth`, and optional `/zedN/pose`. Faster cameras are sampled onto the slowest common timeline there, so they can end up with the same message count as slower cameras. Consumers that care about grouping should follow `/bundle` instead of inferring bundle membership from identical message timestamps.
+
+Use `--bundle-policy strict` when you want thresholded grouping; `--sync-tolerance-ms` only applies in that strict mode. Use `--bundle-policy copy` when you want one MCAP containing all camera namespaces with their original per-camera cadence and no `/bundle` manifest. `copy` disables `--start-frame`, `--end-frame`, and `--sync-tolerance-ms`; `--copy-range common|full` controls whether it trims to the overlap window or preserves each camera’s full timestamp range.
 See [docs/mcap_layout.md](./docs/mcap_layout.md) for the full bundled and single-camera MCAP topic contract.
 
+### MCAP RGBD Viewer
+
+The repo includes an example RGB+depth viewer at `scripts/mcap_rgbd_viewer.py`. It supports standalone `/camera/*` MCAPs, bundled `/bundle` + `/zedN/*` MCAPs, and multi-camera `copy` MCAPs with namespaced `/zedN/*` topics and no `/bundle`.
+
+Install the optional viewer dependencies first:
+
+```bash
+uv sync --extra viewer
+```
+
+Then launch the interactive viewer:
+
+```bash
+uv run --extra viewer python scripts/mcap_rgbd_viewer.py \
+    /workspaces/data/kindergarten/bar/2026-03-18T11-59-41/2026-03-18T11-59-41.mcap \
+    --camera-label zed1
+```
+
+You can also use the same script without a GUI to inspect metadata or render a preview PNG:
+
+```bash
+uv run --extra viewer python scripts/mcap_rgbd_viewer.py \
+    --summary-only \
+    /workspaces/data/kindergarten/bar/2026-03-18T11-59-41/2026-03-18T11-59-41.mcap
+```
+
+```bash
+uv run --extra viewer python scripts/mcap_rgbd_viewer.py \
+    --camera-label zed2 \
+    --frame-index 150 \
+    --export-preview /tmp/mcap_bundled_gap_preview.png \
+    /workspaces/data/kindergarten/throw/2026-03-18T12-58-13/2026-03-18T12-58-13.mcap
+```
+
+The viewer depends on `ffmpeg` being on `PATH` so it can build a seek-friendly preview cache for H.264/H.265 MCAP video streams.
+This is intentionally a simple preview script: it transcodes only the RGB video stream into a temporary intra-frame `mjpeg` cache and then uses that same cache for both scrubbing and normal playback. Depth data is not transcoded to `mjpeg`; it stays in the temporary raw depth cache and is decoded and color-mapped on demand.
+
 ### Why Mixed Hardware/Software Mode Exists
 
 Bundled MCAP export opens one video encoder per camera stream. A four-camera segment therefore consumes four H.264/H.265 encoder sessions at once.