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.

hooks/lib/cli-exit.js

@@ -409,7 +409,7 @@ function runMain(main) {
  *   like `payload`. This exists because `hooks/msd-write-guard.js`'s
  *   emitBlock does NOT write the same bytes to both streams today — it
  *   writes the full JSON `output` to stdout but only the plain-text
- *   `output.reason` STRING to stderr, because Kimi's native hook bus reads
+ *   `output.reason` STRING to stderr, because a native hook bus may read
  *   stderr verbatim back to the model on exit 2. Migrating that call site
  *   onto terminateNow requires a way to say "fd 2 gets this different,
  *   plain-text value" — `stderrPayload` is that seam. Ignored entirely for

hooks/lib/git-cmd.js

@@ -24,8 +24,7 @@
  * §4, epic #3212 Phase 3, #3414) — the built `msd-core/bin/lib/token-scanner.cjs`
  * artifact, not a sibling hooks/-tree file, because hook scripts are staged as
  * standalone files at install time and a sibling require is a staging
- * dependency that can fail silently (see msd-workflow-guard.js's own
- * KIMI_TOOL_NAMES comment for the precedent this follows). Re-exported here
+ * dependency that can fail silently. Re-exported here
  * unchanged — every existing caller's behavior is identical (parity-asserted
  * in tests/token-scanner.test.cjs row 5).
  */

hooks/lib/git-probe.js

@@ -3,8 +3,8 @@
 // distinguishing a GENUINE negative answer (git ran and said "no") from
 // "could not determine" (timeout / spawn failure / signal kill) — #3911.
 //
-// Proven defect: hooks/msd-worktree-path-guard.js, hooks/msd-workflow-guard.js,
-// and hooks/msd-windsurf-pre-write.js each treat spawnSync(git, ..., {timeout})
+// Proven defect: hooks/msd-worktree-path-guard.js and hooks/msd-workflow-guard.js
+// each treat spawnSync(git, ..., {timeout})
 // returning a non-zero/empty result as "not applicable" and allow silently.
 // A macOS CI run showed three deny cases land at 2084ms/2112ms/2177ms — just
 // past the 2000ms budget — each returning exit 0 with EMPTY stdout AND EMPTY

hooks/lib/injection-patterns.js

@@ -12,7 +12,7 @@
  *
  * Why a shared lib require is safe here (the old inlining rationale, retired):
  * the installer stages hooks/lib/ from the MSD_HOOK_LIB_FILES allowlist for the
- * shared-bundle surfaces (Claude-family settings.json runtimes and Kimi), the
+ * shared-bundle surfaces (Claude-family settings.json runtimes), the
  * Cursor stager auto-discovers require('./lib/...') in staged scripts and fails
  * the install loudly when a helper is missing (#2587), and the plugin path
  * ships this directory wholesale.