Correct and trim the comments around the path walks

docs/decisions.md

@@ -325,7 +325,7 @@ Go ties a package to one directory, so folders would split the API into packages
 
 ## The performance gate asserts allocations, not wall-clock time
 
-`AllocsPerRun` is deterministic across machines, so a ±10% ceiling does not flake under
+`AllocsPerRun` is deterministic across machines, so a 10% ceiling does not flake under
 CI load, while time varies with the machine and its neighbours. A rendering slowdown
 almost always costs an allocation too (a lost pre-size, a per-item map, an extra copy).
 The benchmark suite (the README's Development) reports time for a human, not as a pass/fail gate.
@@ -689,8 +689,8 @@ though `hasRead` states the same stop rule. Valid while the two differ in where
 2026-09-28, larv-review on systems-architect's finding; awaiting approval. Goals 2 and
 11: `pathCheck`, `pathCover`, `pathProbe` and `pathDraw` each loop over `stepInto` and
 `table.route` themselves, and `descend` builds its probe in place, so the probe's
-`drawn` map stays on the stack; its caller always builds that map, since no walker
-picks a behaviour by a nil field. One walk calling them through an interface or a type
+`drawn` map stays on the stack; its caller always builds that map, so `readRow` never
+tests it for nil. One walk calling them through an interface or a type
 parameter leaks the walker and the pins it holds, an allocation per reference read.
 Valid while Go's escape analysis treats a call through an interface or a type parameter
 as leaking its receiver.