feat(zed): export depth at neural optimal resolution

Restrict ZED depth export to neural modes and default offline MCAP conversion to the SDK-selected optimal depth size instead of camera resolution.

Add a configurable --depth-size option, propagate the actual returned depth dimensions into DepthMap metadata, and emit a dedicated depth calibration topic when depth resolution differs from video.

Update the batch and recording helper scripts to use the new neural-only depth mode surface and pass through depth sizing.

Verification:
- cmake --build build --target zed_svo_to_mcap mcap_multi_record_tester mcap_pose_record_tester -j4
- build/bin/mcap_multi_record_tester
- build/bin/mcap_pose_record_tester
- build/bin/zed_svo_to_mcap --input /workspaces/data/kindergarten/jump/experiment/2/2026-03-18T11-27-15/2026-03-18T11-27-15_zed4.svo2 --output /tmp/zed4_neural_optimal_test.mcap --codec h264 --encoder-device software --mcap-compression zstd --depth-mode neural --depth-size optimal --start-frame 0 --end-frame 9

scripts/zed_batch_svo_to_mcap.py

@@ -34,6 +34,7 @@ class BatchConfig:
     encoder_device: str
     mcap_compression: str
     depth_mode: str
+    depth_size: str
     with_pose: bool
     pose_config: Path | None
     world_frame_id: str | None
@@ -287,6 +288,8 @@ def command_for_job(job: ConversionJob, config: BatchConfig) -> list[str]:
         config.mcap_compression,
         "--depth-mode",
         config.depth_mode,
+        "--depth-size",
+        config.depth_size,
     ]
     if config.with_pose:
         command.append("--with-pose")
@@ -564,8 +567,14 @@ def run_batch(jobs: list[ConversionJob], config: BatchConfig, jobs_limit: int) -
 )
 @click.option(
     "--depth-mode",
-    type=click.Choice(("neural", "quality", "performance", "ultra")),
-    default="quality",
+    type=click.Choice(("neural_light", "neural", "neural_plus")),
+    default="neural",
+    show_default=True,
+)
+@click.option(
+    "--depth-size",
+    type=str,
+    default="optimal",
     show_default=True,
 )
 @click.option("--with-pose", is_flag=True, help="Enable per-camera positional tracking export when available.")
@@ -615,6 +624,7 @@ def main(
     encoder_device: str,
     mcap_compression: str,
     depth_mode: str,
+    depth_size: str,
     with_pose: bool,
     pose_config: Path | None,
     world_frame_id: str | None,
@@ -638,6 +648,7 @@ def main(
         encoder_device=encoder_device,
         mcap_compression=mcap_compression,
         depth_mode=depth_mode,
+        depth_size=depth_size,
         with_pose=with_pose,
         pose_config=pose_config.expanduser().resolve() if pose_config is not None else None,
         world_frame_id=world_frame_id,

scripts/zed_recording_mcap_tool.py

@@ -66,10 +66,15 @@ def parse_args() -> argparse.Namespace:
     )
     parser.add_argument(
         "--depth-mode",
-        choices=("neural", "quality", "performance", "ultra"),
+        choices=("neural_light", "neural", "neural_plus"),
         default="neural",
         help="Depth mode passed to zed_svo_to_mcap",
     )
+    parser.add_argument(
+        "--depth-size",
+        default="optimal",
+        help="Depth size passed to zed_svo_to_mcap (optimal|native|<width>x<height>)",
+    )
     parser.add_argument("--start-frame", type=int, default=0, help="First SVO frame to convert")
     parser.add_argument("--end-frame", type=int, help="Last SVO frame to convert")
     parser.add_argument(
@@ -289,6 +294,8 @@ def convert_svo(
         args.mcap_compression,
         "--depth-mode",
         args.depth_mode,
+        "--depth-size",
+        args.depth_size,
         "--start-frame",
         str(args.start_frame),
     ]