feat(lidar): add RAVNOVES field review

docs/10_EXTERNAL_PERCEPTION_WORKER.md

@@ -31,6 +31,17 @@ that evidence in Three.js with unrestricted orbit, pan and zoom plus
 intensity/current/candidate/disagreement color modes. The browser never runs
 Patchwork++, parses private firmware or receives a filesystem path.
 
+`missioncore.lidar-field-review/v1` is the separate operator-readable field
+surface. The read-only `/api/v1/lidar/field-reviews` catalog names the exact
+RAVNOVES00 source, timeline and five central urban windows. The bounded
+`/api/v1/lidar/field-reviews/{review_id}/windows/{window_index}` response
+contains a deterministic accumulated map cloud plus masks that were computed
+per original LiDAR sample before accumulation. Its sibling `/preview` endpoint
+returns the content-bound camera JPEG for that window. No source path,
+Patchwork++ binary, worker control or command authority crosses this boundary.
+The legacy E10 source has no intensity, so the API declares it unavailable
+instead of synthesizing a value.
+
 Static K1 3.0.2 firmware evidence confirms that the appliance internally uses
 a Livox MID-360 point/IMU path with richer timestamp/ring semantics and a
 configured MQTT point-cloud downsample factor of four. This creates a concrete

docs/13_LIDAR_WORKER_PRODUCT_AND_ROADMAP.md

@@ -226,6 +226,8 @@ quality line. It is not repaired or hidden by replay.
   source point; raw replay remains unchanged.
 - [x] Add a bounded point-aligned frame API and browser 3D review for
   intensity, current ground, candidate ground and disagreement.
+- [x] Separate the 66-frame indoor contract slice from a named RAVNOVES00
+  field review with five camera-bound urban windows and accumulated map clouds.
 - [x] Record the K1 3.0.2 internal MID-360/LIO field path from static,
   redacted firmware evidence without changing device state.
 - [x] Bind physical height, applied vertical-origin offset and evidence class
@@ -255,6 +257,26 @@ latency remains 0.42 ms. Algorithm IoU is 4.07% p50 and disagreement is 19.13%
 p50. The run is useful for visual review, but its evidence class is
 `operator-estimated`, so input acceptance and production promotion stay false.
 
+The operator-facing field generation is
+`lidar-field-review-57f359dae336f06962e3a29e69e2da1bb8365f9af46d5d3173609f92d43db1ef`.
+It uses the immutable RAVNOVES00 E10 derivative
+`e10-lidar-pack-5da0396d32a27f9d1ca537cc2e8a371d386078d6f0dc71737b78620992af9625`
+over `135.365857292–195.334857292` session seconds. The source has 526
+available LiDAR samples and 1,182,292 points. Five central urban windows cover
+roads, facades, sidewalks, parked vehicles and vegetation. Each window:
+
+- retains the exact source interval and camera preview identity;
+- computes both ground masks independently on every original LiDAR sample;
+- accumulates those point-aligned results only afterwards in the map frame;
+- reports 53–68 source samples and 98,538–172,027 source points;
+- publishes a deterministic uniform display sample of at most 80,000 points.
+
+This field review fixes the operator-context problem of the indoor technical
+slice; it does not upgrade evidence. The E10 derivative remains intensity-free,
+vendor-mapped and without independent ground labels. Its contract is therefore
+`missioncore.lidar-field-review/v1`, `status=diagnostic-only`,
+`decision.status=visual-review-only` and `production_promotion=false`.
+
 The result is an evidence-backed **do-not-promote** decision for the current K1
 feed. Patchwork++ is fast, but its input model assumes a sensor-centric scan
 and physical sensor height. K1 `lio_pcl` is a vendor-mapped increment, its