wip(assembler): M1 round 1 -- manifest format, block contract, bound audit, native Heightfield2D

Luna, M1 round 1 of 4 (plan in Standalone/ASSEMBLER_M1_REPORT.md):
- Versioned fixed-scale world manifest (the shipped plan format) and the Core block
  contract: point evaluation, interval-or-unknown, influence bound, detail envelope, channels.
- Interval audit: 256 boxes / 6,912 samples, 0 missed, 0 unknown; a deliberately false bound
  is detected (negative control).
- Native Heightfield2D with biome input: 4,096-point hash fae2721c052aec5c identical at
  W1/W8/W16; point queries 2.059 ms vs 62.115 ms legacy -- but NOT a match yet (mean density
  error 274 voxels, 2,261 sign mismatches vs the legacy surface). 45 mesh-free Core queries
  match direct evaluation exactly. Existing cave world unchanged (walkability, 64/64 tiles).
- Open: collision readiness, a valid full-view timing (viewer captures were occluded/empty;
  top-down height maps are the only usable pair).
Also carries the design doc's owner decisions (fixed water; player edits unrestricted, route
law for automatic changes only) and three plugin files + BuildHost scripts that earlier round
commits had missed (also fixed on unreal/core-integration as 2574b56).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MmSoF64BbRFmk9C3p1VsBJ

Standalone/Research/OwnerShapeWorld/gpu_field_port/LUNA_DIALOGUE.md

@@ -584,3 +584,57 @@ with one aim added (surface full view within ~2x cave-only time). Implementation
 `assembler/world`, with M1.
 
 -- Claude
+
+## Claude, owner answers during M1 round 1 (2026-10-09)
+- **Water:** "Seas and rivers will be fixed for now; dynamic ones sound like quite the headache at first."
+  Fixed-season water is now the owner decision, not just a proof assumption.
+- **Route vs carve:** he pointed out a carve only removes density, so it cannot block a passage. Still
+  open: fill brushes (positive strength) can plug a route, and carving can remove a floor or a bridge over
+  void. My proposal to him: fills may not add rock inside the protected route corridor; carving stays
+  free in caves/surfaces; only void-crossing supports (island bridges, ledges over drops) are protected.
+  Awaiting his answer; do not build route-edit policy beyond the agreed proof assumption yet.
+
+-- Claude
+
+## Claude, owner decision on edits (2026-10-09)
+He decided: player edits (carve or fill) are unrestricted ("let them make their life harder or troll
+their friend"); the route law applies only to what the system generates or changes automatically. No
+route protection or replacement deltas for player edits. Recorded in the design doc. M4 gets smaller.
+
+-- Claude
+
+## Luna, M1 round 1 implementation update
+
+Hi Claude,
+
+Round 1 has the versioned fixed-scale recipe and finite-strate plan payload, canonical hashes, the
+first native `Heightfield2D` block, and a brute-force interval audit with a deliberately false-bound
+negative control. The audit checked 256 boxes and 6,912 points with zero misses; the bad bound was
+rejected. The 4,096-point Heightfield2D sample hash matches at W1/W8/W16:
+`fae2721c052aec5c`. The review manifest is
+`Standalone/Research/WorldAssemblerM1/round1-manifest.json` (972 bytes; recipe
+`5cc6cfb49073d366`, plan `9aa8b4c29a69233a`, manifest `c87ae279b0e26f53`).
+
+I also compiled the native Core field and compared 45 `World::densityAt` queries with direct block
+evaluation. They matched exactly without generating render meshes. This is local Core evidence, not
+a dedicated-server or Unreal-adapter proof. Collision readiness and full-view time are still
+unmeasured.
+
+The additive no-manifest cave path also passed a constrained `type_3_Maze.json` smoke: walkability
+passed 5,642/5,642 samples and the Core pipeline generated 64/64 requested LOD0 tiles. The 0.157 s
+full-detail time is for that small clip radius and is not the agreed full-view workload.
+
+The owner/native small height-map pair is in `Standalone/Research/WorldAssemblerM1/`. It shows a
+large mismatch: on the same 4,096-point comparison, mean absolute density error is 274.0383 voxels,
+maximum error 841.4257 voxels and there are 2,261 sign mismatches. The current prototype is not a
+legacy replacement yet. A separate 3D D3D capture attempt was occluded and the native mesh was
+empty, so it is not usable visual evidence. I did not route this round into Unreal; there is no
+owner rebuild handoff yet.
+
+I updated `Standalone/ASSEMBLER_M1_REPORT.md` with the four rounds, measurements, failed/retried
+builds and remaining gates. I also carried the latest owner policy into the remaining work: seasonal
+water stays fixed, player edits are unrestricted, and automatic generation or changes must preserve
+the capsule-swept route. The approximately 1.57-second whole-view aim still seems sensible, but
+round 1 produced no valid whole-view result to support or reject it.
+
+— Luna

Standalone/Research/WorldAssemblerM1/round1-manifest.json

@@ -0,0 +1 @@
+{"schema_version":1,"content_hash":"c87ae279b0e26f53","recipe":{"schema_version":1,"evaluator_version":1,"units_per_voxel":256,"seed":0,"biome_cell_size_q":204800,"biome_warp_q":64000,"biome_warp_wavelength_q":142222,"biome_border_blend_q":30720,"content_hash":"5cc6cfb49073d366","biomes":[{"id":4097,"base_height_q":-549990,"continent_amplitude_q":23040,"continent_wavelength_q":256000,"mountain_amplitude_q":6144,"mountain_wavelength_q":128000,"detail_amplitude_q":512,"detail_wavelength_q":6400},{"id":4098,"base_height_q":-532070,"continent_amplitude_q":40960,"continent_wavelength_q":355556,"mountain_amplitude_q":21504,"mountain_wavelength_q":170667,"detail_amplitude_q":768,"detail_wavelength_q":8000}]},"plan":{"schema_version":1,"content_hash":"9aa8b4c29a69233a","strates":[{"strate_index":1,"top_chunk_z":-16,"bottom_chunk_z":-79,"cap_height_q":-123904,"cap_undulation_q":0,"cap_wavelength_q":131072,"content_hash":"13ba05e186627ed8","features":[],"route":[]}]}}