fix(semantics): report the merge-candidate backlog net of scanner noise

The first version of this column reported the raw `merge_candidate_groups`
output, 38, and drew the conclusion that merge candidates are "the one surface
that grew". Measured against the registry, that reading is wrong in both the
number and the direction.

Of the 38 groups, exactly 18 are the Python and TypeScript owner symbols of one
registered vocabulary -- `AGENT_SCOPE_FRONTIER_ACTIONS` with
`AgentScopeFrontierAction`, `EFFECTIVE_ACTIONS` with `EffectiveAction`, and 16
more. A registered cross-runtime vocabulary is *required* to have both ends;
grouping them as merge candidates is the grouping function not consulting the
registry it sits next to. The real backlog is 20, which is smaller than the
historical 32, not larger.

Reporting 38 would have put a scanner artifact into the plan as debt growth --
the exact failure mode this RFC exists to prevent, committed in the column that
claims to measure it. loopx-project#4630 fixes the grouping function itself; this states the
split so the column is right whether or not that lands first.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: song <22676124+songoow@users.noreply.github.com>

Author: songoow

docs/architecture/rfcs/semantic-vocabulary-convergence-v0.md

@@ -870,7 +870,7 @@ vocabulary property the smoke can check. Rows marked *open* wait on a Section
 | Multi-value forks | 4 (1 misclassified) | 4 raw / 3 semantic; `SOURCE_SURFACES` declared as `bounded_context` with four context owners (#4494) | 0 after `scope` declares bounded-context names | M0.5 + baseline PRs |
 | Multi-value twins | 19 | 19, unchanged; 13 after #4617 | 0 | baseline PRs |
 | Legacy should-run fields | 6 fields, 124 py / 10 ts module mentions | 6 fields, 109 py / 10 ts module mentions by identifier counting; below the baseline at every field | 0 fields | M3, identifier-counted |
-| Merge-candidate groups | 32 unreviewed | **38** groups against a baseline of 32; none classified | every group classified; only `same_semantics` groups merged | classification PR, then per-group PRs |
+| Merge-candidate groups | 32 unreviewed | **38** raw groups against a baseline of 32, but 18 of those are exactly the Python and TypeScript owner symbols of one registered vocabulary, so the real backlog is **20** and it is smaller than the baseline, not larger; none classified | every group classified; only `same_semantics` groups merged | classification PR, then per-group PRs |
 | Control-plane py/ts twins | 43 | 43, unchanged | follows the TypeScript migration RFC; no target here | M4 |
 
 The *Measured* column is dated and non-normative: it records what the tree

docs/architecture/rfcs/semantic-vocabulary-convergence-v0.zh-CN.md

@@ -715,7 +715,7 @@ TypeScript effective-action 绑定与[术语表](../../reference/glossary.md)通
 | 多值分叉 | 4（1 个误分类） | 原始 4 / 语义 3；`SOURCE_SURFACES` 已声明为 `bounded_context`，带四个上下文 owner（#4494） | `scope` 声明有界上下文名字后为 0 | M0.5 + 基线窄 PR |
 | 多值孪生 | 19 | 19，未变；#4617 合并后为 13 | 0 | 基线窄 PR |
 | 旧 should-run 字段 | 6 个字段，124 py / 10 ts 模块提及 | 6 个字段，按标识符计数为 109 py / 10 ts 模块提及；每个字段都低于基线 | 0 个字段 | M3，按标识符计数 |
-| 合并候选组 | 32 组未评审 | **38** 组（基线为 32）；尚无一组完成分类 | 每组已分类；只合并 `same_semantics` 的组 | 分类表 PR，随后逐组 PR |
+| 合并候选组 | 32 组未评审 | 原始 **38** 组（基线为 32），但其中 18 组正是同一个注册词表的 Python 与 TypeScript owner 符号，真实待分类为 **20** 组——比基线更少而非更多；尚无一组完成分类 | 每组已分类；只合并 `same_semantics` 的组 | 分类表 PR，随后逐组 PR |
 | 控制面 py/ts 孪生 | 43 | 43，未变 | 跟随 TypeScript 迁移 RFC；本 RFC 不设目标 | M4 |
 
 *实测* 列带日期，且**非规范**：它记录该日期的树上实测值，使计划可以按当前数字