chore(release): v6.0.0 — "Prose is not a contract" (MAJOR: finding IDs name the check)

MAJOR, because M-BUG-28 changed what a finding ID means: the {NNN} in
CA-{SCANNER}-{NNN} names the CHECK, not its emission position. IDs are
therefore not unique per finding -- one check failing in three files emits
three findings sharing an ID -- and any consumer keying on `id` alone must
move to (id, file, line).

The release theme is a class of defect rather than a feature area. Three
sweeps (Q1, Q2, Q_AUDIT) kept surfacing the same shape: a command template
stated an invariant in prose, code on the other side depended on it, and
nothing checked that the two still agreed. Q1 put the write-scope gate in
code (measured: 9 writers under scanners/, 1 imported the gate). Q2 checks
each template's argv against the CLI that receives it (--stale-after 30
reached its CLI as one argument under zsh and was ignored while the command
reported success). Q_AUDIT measured and ranked the third instance --
data contracts hand-built by the model and parsed by engines that know one
frozen example -- with the recovery path on top, deliberately not yet fixed.

Measured this session, not carried forward:
  37 commits since v5.13.0 (git log v5.13.0..HEAD --oneline | wc -l)
  1752 tests, 0 failing, post-bump
  self-audit --check-readme: PASS, readmeCheck.passed, 0 mismatches
  check-versions.mjs: 0 ERROR (1 WARN = the unreleased bump this cut closes)
  scanners 16, agents 7, commands 21, hooks 4, knowledge 8

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XKeJyEVn9GuBcBQbE6EYwm

.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "config-audit",
   "description": "Multi-agent workflow for analyzing, reporting, and optimizing Claude Code configuration across your entire machine",
-  "version": "5.13.0",
+  "version": "6.0.0",
   "author": {
     "name": "Kjell Tore Guttormsen"
   },