feat(3347): opt-in auto-update of knowledge graph after main HEAD advances

Closes #3347

Config:
- Add graphify.auto_update (default false) to manifests:
  sdk/shared/config-defaults.manifest.json, config-schema.manifest.json

Hook:
- hooks/gsd-graphify-update.sh — PostToolUse Bash matcher
  - Gates: tool_name=Bash, HEAD-advancing git op, CI=unset, in git repo,
    current branch == default branch (git.base_branch override or main/
    master/trunk fallback), graphify.enabled && graphify.auto_update both
    true, graphify on PATH, no live PID lock
  - Writes .planning/graphs/.last-build-status.json with status=running
    synchronously, then detaches hooks/lib/gsd-graphify-rebuild.sh
- hooks/lib/gsd-graphify-rebuild.sh — detached rebuild runner
  - PID-lock acquire + trap-on-exit cleanup
  - graphify update . then cp graphify-out/* → .planning/graphs/
  - Status file rewritten to status=ok|failed with exit_code, duration_ms,
    head_at_build
- Portable detach (subshell + disown, no setsid dependency)

Installer:
- bin/install.js: register hook as PostToolUse Bash matcher (5s timeout)
- Add to gsdHooks uninstall list and expectedShHooks warning list

Planner / researcher status surface (issue #3347 reviewer must-have AC):
- agents/gsd-planner.md and agents/gsd-phase-researcher.md
  load_graph_context steps now read .last-build-status.json and surface:
  running → "rebuild in flight"; failed → "auto-rebuild FAILED at {ts},
  context is from prior build"; ok with stale head_at_build → "HEAD has
  advanced since last build"

Settings:
- get-shit-done/workflows/settings.md adds "Graph auto-update" question
  with No-Recommended default; bullets and update_config block updated

Inventory:
- docs/INVENTORY.md hook count 12 → 13 with new row
- docs/INVENTORY-MANIFEST.json regenerated

Tests:
- tests/feat-3347-graphify-auto-update-config.test.cjs (8 tests):
  isValidConfigKey accepts graphify.auto_update, CANONICAL_CONFIG_DEFAULTS
  default false, config-set round-trip, sibling key preservation
- tests/feat-3347-graphify-auto-update-hook.test.cjs (18 tests):
  all bail paths (non-Bash, non-HEAD-advancing, enabled=false,
  auto_update=false, CI=true, non-default-branch, missing graphify bin,
  live-PID lock), dispatch path with mock graphify bin (sync running
  status + detached transition to ok/failed), stale-PID lock, all five
  HEAD-advancing command matchers, git.base_branch override

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

.changeset/3347-graphify-auto-update-hook.md

@@ -0,0 +1,6 @@
+---
+type: Added
+pr: 0
+---
+
+**Opt-in: auto-rebuild knowledge graph after main HEAD advances** — new config key `graphify.auto_update` (default `false`) and bundled PostToolUse hook `hooks/gsd-graphify-update.sh` keep the `.planning/graphs/graph.json` consumed by `gsd-planner` and `gsd-phase-researcher` current without manual `/gsd:graphify build` runs. When both `graphify.enabled` and `graphify.auto_update` are `true`, the hook fires after Bash tool calls matching HEAD-advancing git ops (`commit`, `merge`, `pull`, `rebase --continue`, `cherry-pick`) on the default branch, writes a synchronous `running` status to `.planning/graphs/.last-build-status.json`, then dispatches `graphify update .` in a detached subprocess that updates the status file to `ok` (with `duration_ms` + `head_at_build`) or `failed` (with `exit_code`). The planner and researcher's `load_graph_context` steps now surface the auto-build state alongside the existing staleness annotation — including the must-have failure-surface case from the issue review (`"auto-rebuild FAILED at {ts}; context is from the prior build"`). PID-locked against concurrent rebuilds (stale-PID tolerant via `kill -0`), CI-aware (`$CI` env suppresses), and bails silently if `graphify` is not on `PATH` or the current branch is not the default. `/gsd:settings` adds a "Graph auto-update" question gated on Graphify being enabled. Closes #3347.