refactor: drop 12 runtimes, keep Claude, Codex, OpenCode, Cursor, ZCode, Antigravity

Removes kilo, kimi, kimi-code, copilot, windsurf, augment, trae, qwen, hermes,
cline, codebuddy and pi end to end: capability descriptors, installer branches
and converters (bin/install.js 14.9k -> 11.2k lines), TypeScript converters,
hook surfaces and runtime homes, review lanes qwen/kimi-code, the two pi
migrations, Kimi payload normalization in the hook guards, dead hostBehaviors
vocabulary, launcher home probes, fixtures, runtime-specific tests and the
prose that presented them as supported.

Installer output for the six kept runtimes is byte-identical to before the
prune. The Kimi tool-vocabulary tests in workflow-guard, read-guard and
read-injection-scanner are left in place pending a decision.

scripts/prompt-injection-scan.sh

@@ -137,10 +137,6 @@ ALLOWLIST=(
   # ("ignore previous instructions", "you are now…") as examples agents must
   # NOT comply with — it is the defense, not an attack vector.
   'references/untrusted-input-boundary.md'
-  # Security regression tests for input validators — fixtures must contain
-  # real injection payloads to prove the validator rejects them. See
-  # DEFECT.PROMPT-INJECTION-SCAN-COLLISION in CONTEXT.md.
-  'tests/windsurf-conversion.test.cjs'
   # RuleTester fixtures for the local/no-unguarded-nonportable-exec ESLint rule
   # contain shell-exec command strings (exec("sh -c …"), execFileSync('bash',['-c',…]))
   # as test DATA the rule must lint — not attack vectors. ADR-1703 Phase 3 (#1720).
@@ -149,12 +145,6 @@ ALLOWLIST=(
   # exec command strings (execFileSync('npm', ['install'])) as test DATA the rule
   # must lint — not attack vectors. ADR-1703 Phase 4 (#1726).
   'tests/no-bare-npm-exec.rule.test.cjs'
-  # #2547 — the Kimi field-shadowing regression proves msd-prompt-guard still
-  # SCANS the reconstructed edit[].new content when a model-supplied new_string
-  # tries to shadow it. The fixture must be a real injection phrase or the test
-  # asserts nothing: it is the payload the guard is required to catch, carried
-  # as test DATA. Same class as the read-injection-scanner suites above.
-  'tests/kimi-payload-field-shadowing.security.test.cjs'
   # Phase-ID grammar regression tests exercise `RegExp.prototype.exec` via
   # `re.exec('<phase-id>')` against fixtures like 'MANIFOLD-64-auth' / 'CK-64-auth'.
   # The scanner's `exec('` code-execution pattern matches that benign method call,