Tighten Tier2 stock decode boundary

docs/control-loop-atlas/runtime-roots-camera-and-support-families.md

@@ -1426,6 +1426,15 @@
   tail-calling `0x00419230`. So the remaining Tier-2 source problem is increasingly “which stock
   rows that rebuild admits or seeds with nonzero bank bytes” rather than “which unrelated later
   service invokes the banked clone pass.”
+  The source-decoder edge inside that handoff is tighter in the same direction too. Direct
+  disassembly of `0x00414490` shows it already using the imported stock tail structurally rather
+  than merely forwarding it: after importing `[record+0xb8/+0xb9/+0xba/+0xbb]`, it derives the
+  optional-plane byte count from `[record+0xb8] * [record+0xb9] << 5`, allocates and zeroes the
+  four optional plane buffers at `[record+0xcf/+0xd3/+0xd7/+0xdb]`, and then uses the high nibble
+  of `[record+0xba]` while decoding those packed planes. So the remaining Tier-2 seam is not below
+  “raw stock bytes exist at all”; it is above a source decoder that already materializes the
+  shape-plane state before the later `0x00416ce0` bare-name remap and `0x00419230` rebank-or-clone
+  pass.
   The stock owner chain above those parser fields is explicit now too:
   `0x00438c8e -> 0x004131f0 -> 0x00412fb0 -> 0x004120b0` constructs root `0x0062b268`, and the
   adjacent `.rdata` strings at `0x005c93f4..0x005c940e` prove that `0x00412fb0` is the