Narrow engine type auxiliary stem frontier

docs/rehost-queue.md

@@ -12,16 +12,19 @@ This file is the short active queue for the current runtime and reverse-engineer
 
 - The active static parser head is now the `engine_types` semantics frontier.
   The repo now has structural inspectors for `.car`, `.lco`, `.cgo`, and `.cct`, but the binary side is still only partially semantic: the checked 1.05 corpus grounds `.car` fixed strings at `0x0c / 0x48 / 0x84` plus a second fixed stem slot at `0xa2` and a side-view resource name at `0xc0`, while `.lco` carries a stable primary stem at `0x04` and only conditional companion/body slots at `0x0c` and `0x12` when the leading stem slot is padded.
-  The checked 1.05 corpus now also splits `.car` auxiliary stems into `126` direct matches, `14` role-neutral roots, and only `5` truly distinct cases, with those five exact internal-to-auxiliary pairs now preserved directly in the report surface, while `.cgo` collapses into five stable scalar ladders instead of arbitrary floats.
+  The checked 1.05 corpus now also splits `.car` auxiliary stems into `126` direct matches, `14` role-neutral roots, and only `5` truly distinct cases, with those five exact internal-to-auxiliary pairs now preserved directly in the report surface.
+  That distinct cohort is narrower than it first looked: `3` of the `5` are just opposite-role tender-to-loco pairings (`ClassA1T -> ClassA1L`, `CramptonT -> CramptonL`, `WhaleT -> WhaleL`), and only the two `QJ` rows remain as reordered alias roots.
+  On the cargo side, `.cgo` now collapses into five stable scalar ladders instead of arbitrary floats.
   The early `.lco` lane block is now partially partitioned too: only offsets `0x20`, `0x34`, `0x38`, `0x3c`, `0x44`, `0x48`, and `0x54` behave like low-cardinality buckets, while the other early lanes still look high-variance.
   The side-view resource path is now grounded into `Data/2D/rt3_2IMB.PK4`, and the `.imb` parser now decodes shipped comment-suffixed numeric rows plus `_NE` profile fields such as `HorizontalScaleModifier` and `ImageWHScaled`.
   The checked PK4 linkage split is now explicit too: `132 / 145` side-view resource names resolve directly in 1.05, but the remaining `13` are the missing `CarSideView_3.imb` cohort and that hole exists in both checked installs, while `43 / 145` derived `{internal_stem}_NE.imb` names resolve and all of those hits belong to matched locomotive pairs.
   The parser now preserves that `CarSideView_3` miss cohort exactly, and it also preserves the tiny conditional `.lco` companion-profile seam directly: in 1.05 the padded `.lco` companion/body slots collapse to `Zephyr / (none) / CarSideView_3` for `242_a1_l`, `gp35l`, `u1l`, and `zephyrl`, and to `VL80T / Loco / CarSideView_{1,2}` for `be 5-7`, `f3 loco`, and `gp7`.
   The classic install widens the same `0xc0` seam further: the side-view slot is not only `CarSideView_*` but also a larger unresolved `*_Profile.imb` family, and every checked classic `*_Profile.imb` reference currently misses the packaged `rt3_2IMB.PK4` surface too.
+  A whole-tree file census now rules out the simple fallback too: none of those `*_Profile.imb` names, and no `CarSideView_3.imb`, exist as loose files anywhere in the checked `rt3/` or `rt3_105/` trees.
   The packaged profile metadata is stable enough to summarize: `CarSideView_1` is `512x512` at `0.04` VRAM, `CarSideView_2` is `512x256` at `0.02`, and every packaged `_NE` profile is `512x128` with `HorizontalScaleModifier = 0.75` and `MaxPercentOfInterfaceVRAM = 0.09`.
   The `_NE` split is now aligned with the locomotive display census too: all `43` packaged `_NE` hits live inside the grounded display prefix, and all `5` unmatched display-tail families are still missing packaged `_NE` profiles.
   The cargo side is partially linked now as well: the `.cgo` ladder families and `.cct` sidecar identifiers share the same cargo-family keys for ten checked families, with `Troop` left as the only `.cct`-only outlier.
-  The next honest static work is to keep promoting those fixed lanes into stable parser fields, explain the five remaining distinct auxiliary-stem cases, decide whether the classic `*_Profile.imb` side-view references are dead loose-file dependencies or a still-unmapped package family, and decide how far the `.cgo` ladders plus the low-cardinality `.lco` lanes can be grounded without overclaiming semantics. The latest corpus check did narrow one point already: the low-cardinality `.lco` lanes do not split cleanly on `_NE` presence, so that branch now wants binary/code correlation rather than more aggregate-only counting.
+  The next honest static work is to decide whether the two remaining reordered `QJ` auxiliary roots are just alternate content aliases or evidence of a narrower foreign-display/image root, decide whether the classic `*_Profile.imb` side-view references are dead loose-file dependencies or a still-unmapped package family, and decide how far the `.cgo` ladders plus the low-cardinality `.lco` lanes can be grounded without overclaiming semantics. The latest corpus check did narrow one point already: the low-cardinality `.lco` lanes do not split cleanly on `_NE` presence, so that branch now wants binary/code correlation rather than more aggregate-only counting.
   Preserved checked parser detail now lives in [EngineTypes parser semantics](rehost-queue/engine-types-parser-semantics-2026-04-21.md).
   Preserved checked format inventory detail now lives in [RT3 format inventory](rehost-queue/format-inventory-2026-04-21.md).
 

docs/rehost-queue/engine-types-parser-semantics-2026-04-21.md

@@ -23,6 +23,14 @@ first `.car` / `.lco` / `.cgo` / `.cct` inspector pass landed.
   - `WhaleT -> WhaleL`
   - `classqjl -> qjclassl`
   - `classqjt -> qjclasst`
+- The parser now preserves one tighter split inside those five distinct cases:
+  - `paired_opposite_role_suffix`: `3`
+    - `ClassA1T -> ClassA1L` on `class_a1t`
+    - `CramptonT -> CramptonL` on `crampton tender`
+    - `WhaleT -> WhaleL` on `orca nx462 tender`
+  - reordered `QJ` aliases: `2`
+    - `classqjl -> qjclassl` on `class_qjl`
+    - `classqjt -> qjclasst` on `class_qjt`
 - `.lco` carries one always-present primary stem at `0x04`.
 - `.lco` only carries meaningful secondary slots when that leading stem slot is padded:
   - `0x0c`: conditional companion stem such as `VL80T` or `Zephyr`
@@ -119,6 +127,10 @@ first `.car` / `.lco` / `.cgo` / `.cct` inspector pass landed.
   - `TenWheeler_Profile.imb`
   - `V2Class_Profile.imb`
   - `Vittorio_Profile.imb`
+- A whole-tree file census now rules out the simplest fallback for that family:
+  - none of those checked classic `*_Profile.imb` names exists as a loose file anywhere under the
+    checked `rt3/` or `rt3_105/` trees
+  - `CarSideView_3.imb` also does not exist as a loose file anywhere under those trees
 - The packaged profile metadata is now bounded too:
   - `CarSideView_1.imb`: `512x512`, `MaxPercentOfInterfaceVRAM = 0.04`
   - `CarSideView_2.imb`: `512x256`, `MaxPercentOfInterfaceVRAM = 0.02`
@@ -168,8 +180,10 @@ first `.car` / `.lco` / `.cgo` / `.cct` inspector pass landed.
 ## Remaining Static Questions
 
 - `.car`
-  - what the `0xa2` auxiliary stem really represents in the five remaining distinct cases:
+  - what the `0xa2` auxiliary stem really represents in the two remaining reordered `QJ` cases:
     alternate content root, paired tender/loco image root, or a narrower foreign-display alias
+  - whether the three opposite-role pairings are best described as direct tender-to-loco image
+    roots or some slightly broader paired-display fallback convention
   - whether the trailing side-view resource can be tied cleanly to the PK4-backed `CarSideView_*`
     metadata, the classic `*_Profile.imb` reference family, and the engine-specific `_NE.imb`
     profiles without inventing frontend semantics