Answer prose round 1: state the gate once in item 5, title the engine decision by what this commit holds, and plainer wording across items 97 to 107 and the decisions

docs/decisions.md

@@ -504,10 +504,10 @@ option earns its place.
 
 A link is proved against the parent's keys and a key's uniqueness is a data mistake, so
 both are load-time; the row lookup, by name and by parent, serves only a draw, so it waits
-for the first one, keeping `New` linear in the bytes read. On 2026-10-05 building every
-shipped lookup at load took 13 ms and 3.2 MiB on top of a 60 ms `New`, a cost that grows
-with every locale and with item 78 while most tables are never drawn by name or parent;
-the maintainer kept the lookup on first draw, written once.
+for the first one, keeping `New` linear in the bytes read. On 2026-10-05, building every
+shipped lookup at load added 13 ms and 3.2 MiB to a 60 ms `New`. That cost grows with every
+locale and with item 78, and most tables are never drawn by name or parent. The maintainer
+kept building the lookup on first draw, once per table.
 
 ## Two categories may name one TSV
 
@@ -827,15 +827,16 @@ Valid while names are read only inside the category binding them.
 
 ## comprehension floor: every dimension and the overall at 7.0 or above; the change making every `{…}` draw afresh and items 97 to 106 ship with no scoring run, and item 5 scores them
 
-2026-10-04, Lilleman auf Larv. Serves goal 3. The nine-seat panels in `todo.md`'s git history, 2026-09-22 to 2026-09-30, held the score between 5.2 and 6.1, with Locality lowest at every seat, and the hardest units were the draw fences that change deleted. The change making every `{…}` draw afresh and `todo.md` items 97 to 106 delete or move what the earlier chunks of the below-floor round touched, so a scoring run before item 106 merges would score code about to go; the calc change accepting a name as an operand shipped unscored on 2026-10-05 for that reason. They finish the round `AGENTS.md`'s comprehension rule lets merge below the floor, a restructure the maintainer approved. No scoring run or panel runs until item 5, which runs the nine-seat panel and larv-review's scoring run. The scoring run writes this entry's first scores. Valid until `todo.md` item 5 runs.
+2026-10-04, Lilleman auf Larv. Serves goal 3. The nine-seat panels in `todo.md`'s git history, 2026-09-22 to 2026-09-30, held the score between 5.2 and 6.1, with Locality lowest at every seat, and the hardest units were the draw fences that change deleted. The change making every `{…}` draw afresh and `todo.md` items 97 to 106 delete or move what the earlier chunks of the below-floor round touched, so a scoring run before item 106 merges would score code about to go; the calc change accepting a name as an operand shipped unscored on 2026-10-05 for that reason. With that change, items 97 to 106 are the restructure the maintainer approved: the round `AGENTS.md`'s comprehension rule lets merge below the floor. No scoring run or panel runs until item 5, which runs the nine-seat panel and larv-review's scoring run. The scoring run writes this entry's first scores. Valid until `todo.md` item 5 runs.
 
-## The template engine stays the root package until item 107, and what stands alone lives in `internal/`
+## The template engine stays the root package until item 107, and items 97 to 103 move what stands alone into `internal/`
 
 2026-10-05, Lilleman auf Larv. Serves goals 3.2 and 3.4. Two architect reviews found the
 engine's types, `template`, `table`, `arm`, `op` and `nameBinding`, bound in one cycle because several passes each write part of the same structs; the template
-language's recursion needs only a little of it. Moving the engine before items 101 to 106
-give each value one writer would spread that knot across packages, behind forwarding calls
+language's recursion reads few of those types. Moving the engine before items 101 and 104 to
+106 give each value one writer would spread that knot across packages, behind forwarding calls
 from every public method. The draw state and `DataType` read no engine type, and `DataType`
-moves with the proofs' bounds keyed by it; the grammar, the reading of data files, the
-proofs' bounds, the builtins and a table's rows stop reading one once items 98 to 103 cut
-their edges, so they move now, each behind an import list a test checks. Valid until `todo.md` item 107 moves the engine.
+moves with `proven`, what a proof knows of a value, which is keyed by it; the grammar, the
+reading of data files, `proven`, the builtins and a table's rows stop reading any engine
+type once items 98 to 103 remove those reads, so they move now, each behind an import list
+a test checks. Valid until `todo.md` item 107 moves the engine.