Resolves 5 P1 review threads all reporting the same class of issue:
inline-code spans (backticks) and URLs broken across newlines.

CommonMark inline-code spans cannot span newlines — the `span` is
literally broken as rendered, and readers cannot copy the path
cleanly. Same for URLs: Markdown auto-linkers stop at whitespace.

Fix pattern: move the full backticked path (or URL) onto its own
line, wrapping the surrounding prose instead. No content removed.

Threads addressed:
- 59Vtvx — OpenAI Frontier URL split across line break
- 59WtwY — `docs/research/frontier-ux-zora-evolution-*.md`
  split (first occurrence, Class-a list)
- 59Wtwq — `memory/feedback_aaron_dont_wait_on_approval_log_
  decisions_frontier_ui_is_his_review_surface_*.md` split;
  also updated to the concrete landed filename
- 59Wtw8 — same UX design-doc path split (Composition section)
- 59WtxM — `docs/definitions/KSK.md` split

Same fix pattern as PR #352 (server-meshing-*.md path).

…361)

* fix: strip history-log commentary from TemporalCoordinationDetection

Aaron's correction: "comments should not read like history, what
use is this to a future maintainer? Code comments should explain
the code not read like some history log ... code is not a history
file ... there should be existing lint hygiene for that."

GOVERNANCE §2 + CLAUDE.md both say the same thing: docs read as
current state, not history. This file had 27 occurrences of
graduation / ferry / attribution / Otto-NNN / "Provenance:" /
"per correction #N" spread across the module header and six
function docs. Every occurrence is process metadata — who wrote
it, which round shipped it, which absorb it traces to — none of
which a future maintainer reading the function body cares about.

Kept (explains code):
  * math: ranges, Euler identity, atan2 + IEEE-754 signed zero,
    epsilon-floor rationale
  * complementarity: PLV vs cross-correlation, magnitude vs
    offset, same-phase vs anti-phase distinction
  * input contracts: None semantics, radians, wrapping-
    convention-agnostic
  * composition guidance: prefer phaseLockingWithOffset for
    single-pass; burstAlignment is pair-wise, node-set
    generalisation belongs elsewhere

Removed (history):
  * module-header "Aaron's design / Amara's formalization /
    Attribution / Scope of this first graduation"
  * per-function "Provenance: ... N-th courier ferry ...
    Nth graduation under the Otto-105 cadence"
  * "Per Amara 18th-ferry correction #6" prefix (the argument
    it introduced is preserved verbatim, just without the prefix)
  * broken inline-code span across newlines in the removed
    ferry-path (same class as #348's fix)

37 TCD tests pass. Build clean, 0 warnings. No code bodies
changed — this is a doc-comment-only cleanup. Net -64 lines.

Follow-up: factory-wide audit + lint for the same class of
violation (row landing next tick per queue-saturation
discipline).

* fix(#361): 3 review threads — module-doc accuracy + PLV wording

Thread 59Y_Np (Codex P2) + thread 59Y_st (Copilot P1) — same
finding: the module-level summary claimed "All entry points
return Option-wrapped values" and listed "length-mismatched"
as a None condition, but:
  * crossCorrelationProfile / significantLags / burstAlignment
    return plain arrays, not Option-wrapped
  * crossCorrelation tolerates mismatched lengths — it operates
    over the overlap window at the given lag

Split the module doc into two return-shape families (single-
value primitives that return Option, profile/array primitives
that return arrays) and added an explicit note on length
semantics: crossCorrelation tolerates mismatch, phase-pair
primitives (PLV / meanPhaseOffset / phaseLockingWithOffset)
require equal lengths.

Thread 59Y_tT (Copilot P2) — meanPhaseOffset doc wording:
"vector whose magnitude is phaseLockingValue" implied the
function IS the magnitude. Corrected to "whose magnitude is
the PLV (i.e. the value returned by phaseLockingValue on the
same inputs)".

All three are doc-comment-only corrections. 37 TCD tests pass;
build clean.

)

* loop: tick-history row — Otto-219..221 drain + Otto-220 correction

Three-PR drain (#348 clean, #340 merged, #361 opened) plus
Aaron's Otto-220 correction on code-comments-vs-history
absorbed into a durable memory + PR #361 fix landed against
main.

No code changes; append-only history row per
docs/AUTONOMOUS-LOOP.md end-of-tick checklist.

* fix(#362): typo "don not integrate" -> "don't integrate"

Copilot P2 thread 59ZEMT — one-character typo fix in the
quoted-text portion of the Otto-219..221 tick row's Observation
4. The surrounding quote survives; only the typo shifts.