docs(accuracysnes): C11.12 is unauthorable — verified, not assumed

CHANGELOG.md

@@ -11,6 +11,22 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- **`C11.12` is unauthorable, verified two ways — there is nothing to assert.** Checked rather than
+  assumed: the row's subject ("Mode 7 scroll offset latch timing") appears exactly **twice** in
+  `ref-docs/`, both times as a game-compatibility table row naming NHL '94, with **no behavioural
+  statement anywhere** — no value, no timing, no mechanism. And no reference models a distinct
+  timing: ares reads the register live in `sfc/ppu/mode7.cpp`
+  (`s32 hoffset = (i13)self.io.hoffsetMode7;`), with no per-scanline latch for a test to catch a core
+  getting wrong.
+
+  Writing a row would mean inventing an expectation and checking our own arithmetic against itself —
+  the `E9.11` failure mode. The genuinely testable part of `$210D`'s dual-latch behaviour (it updates
+  `M7HOFS` via `M7_old` *and* `BG1HOFS` via `BG_old`, with different formulas) is **already covered**
+  by `C4.01` and `C4.04`/`C4.05`, all blessed scenes; `C11.12` adds only the word *timing*, which is
+  the part nothing defines.
+
+  It belongs with `C13.*` and `F1.22`: enumerated, uncoverable, scored as such rather than chased.
+
 - **`inidisp_forgot_to_force_blank`: the model difference is now identified exactly, and a recorded
   claim about it is corrected.** The last per-dot framebuffer gap was documented only as "an
   `internal_cgram_address` draw-ordering detail". The concrete difference: MesenCE updates

docs/accuracysnes-plan.md

@@ -766,6 +766,36 @@ wrapped part is page zero and precisely page zero. Nothing else is in reach howe
 which is what lets the program wait a whole buffer cycle instead of timing anything — and the echo
 offset free-runs regardless of `FLG` bit 5, so there is no starting phase to know.
 
+### `C11.12` is unauthorable — there is nothing to assert, not merely something hard
+
+Checked 2026-08-02, two ways, and recorded so it is not re-attempted.
+
+**The corpus gives a game name and nothing else.** `C11.12` ("Mode 7 scroll offset latch timing")
+appears exactly twice in `ref-docs/`, both times as a row in a game-compatibility table naming
+**NHL '94**. There is no behavioural statement anywhere — no value, no timing, no mechanism. A test
+needs something to assert, and the research corpus supplies nothing to assert.
+
+**No reference models a distinct timing either.** ares reads the register **live** in
+`sfc/ppu/mode7.cpp`:
+
+```cpp
+s32 hoffset = (i13)self.io.hoffsetMode7;
+```
+
+There is no per-scanline latch, no deferred copy, nothing a test could catch a core getting wrong.
+Writing a row would mean inventing an expectation and then checking that our own arithmetic agrees
+with itself — the `E9.11` failure mode.
+
+**What the dual-latch mechanism actually is, and where it is already covered.** fullsnes documents
+that `$210D` updates `M7HOFS` (via `M7_old`) *and* `BG1HOFS` (via `BG_old`), with different
+formulas. That is real and testable — and it is already covered: `C4.01` (the BG formula keeps the
+low three bits) and `C4.04`/`C4.05` (`$210D` drives Mode 7 scroll) are both blessed scenes. `C11.12`
+adds only the word *timing*, which is the part nothing defines.
+
+**To ever author it** would need a source that says what NHL '94 depends on — a disassembly note, a
+hardware measurement, or a reference that special-cases it. Until then it belongs with `C13.*` and
+`F1.22`: enumerated, uncoverable, and scored as such rather than chased.
+
 ### Survey of what is left in Group E, and why none of it is a quick row
 
 Six rows landed in this batch (`E8.01`, `E9.02`, `E5.06`, `E9.09`, `E3.09`, `E3.13`) at roughly one