annotate: address review feedback — bug fixes, docs/code drift, naming, cleanup

Bugs
  * validator: don't re-raise on unknown style. The second column_for_style
    lookup (used to route persistent vs event) now sits in try/except so an
    unknown style is recorded by _check_column_routing and skipped instead
    of crashing the whole validation pass.
  * general_vqa._target_cameras: when restrict_to_default_camera is set but
    the configured camera_key isn't one the provider exposes, warn and fall
    back to all cameras instead of returning a phantom key that KeyErrors
    deep in frame decode.
  * interjections: clamp interjection timestamps to frame_timestamps[0]
    rather than a hardcoded 0.0 (datasets can start at non-zero t).

Docs / code drift
  * annotation_pipeline.mdx: drop the phantom 'vocabulary discovery / phase
    0 / --vocabulary.* / canonical_vocabulary.json' section (none of it
    exists); describe the real describe->segment + coverage-stitch flow.
    Soften the src/lerobot/tools/ + TOOL_REGISTRY reference to 'not part of
    this PR' (matches tools.mdx, which already marks the runtime layer as
    not-yet-implemented). Fix the --push_to_hub/--new_repo_id wording. Note
    the default is now a single h200. Add a 'Contributing new modules'
    section inviting module / prompt / quality contributions.
  * executor docstring: six phases, no phantom phase 0.

run_hf_job.py
  * add the Apache 2.0 license header (was flagged repeatedly).
  * default to a single GPU: flavor=h200, parallel_servers=1, num_gpus=1
    (scale to h200x4 noted in the docstring).
  * pin the install to @main instead of the feature branch (won't break
    after merge).

Naming / cleanup
  * rename dest_repo_id -> new_repo_id across config / script / example /
    test to match the LeRobot dataset edit tools.
  * rename prompt templates module_N_*.txt -> descriptive (plan_*,
    interjections_*, vqa.txt) and update every load_prompt() call.
  * remove dead _messages_to_prompt (used only by the removed in-process
    backends).
  * declare _warned_decode_fail (frames) and _warned_no_camera (vqa) as
    real init=False dataclass fields instead of getattr monkey-patches.
  * scope bandit B607 to the two ffmpeg subprocess.run sites via
    '# nosec B607' and drop it from the global skip list.

Tests
  * fix stale canned-VLM markers ('ONE realistic interruption' ->
    'compact interjection', 'Update the memory' -> 'compressed semantic
    memory') and drop the dead 'concise hierarchical PLAN' plan responders
    (plan generation is deterministic now) in run_e2e_smoke,
    test_pipeline_recipe_render, test_modules.
  * run_e2e_smoke now asserts interjection + speech rows are produced so a
    stale marker can't silently pass again.
  * drop remaining 'PR 1' / 'PR 2' references from test comments / names.

Verified: tests/annotations + tests/datasets/test_language +
tests/scripts/test_lerobot_annotate (31 passed); make-style E2E smoke
(interjections=1 speech_atoms=2); pre-commit (ruff, mypy, bandit,
prettier) clean.

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

tests/annotations/run_e2e_smoke.py

@@ -60,13 +60,11 @@ def _stub_responder(messages):
                 {"text": "place the bottle down", "start": 2.0, "end": 3.0},
             ]
         }
-    if "concise hierarchical PLAN" in text:
-        return {"plan": "1. grasp\n2. pour\n3. place"}
-    if "Update the memory" in text:
+    if "compressed semantic memory" in text:
         return {"memory": "poured once"}
     if "acknowledgement the robot" in text:
         return {"text": "Sure."}
-    if "ONE realistic interruption" in text:
+    if "compact interjection" in text:
         return {"interjection": "use less water", "speech": "Using less water."}
     if "frame-grounded visual question" in text:
         return {"question": "How many cups?", "answer": {"label": "cup", "count": 1}}
@@ -94,6 +92,23 @@ def main() -> int:
         print(f"phases={[(p.name, p.episodes_processed) for p in summary.phases]}")
         print(f"validation: {summary.validation_report.summary()}")
         print(f"shards rewritten: {len(summary.written_paths)}")
+
+        # Assert the interjection code path actually fired — otherwise a stale
+        # canned-VLM marker would silently produce zero interjections and this
+        # smoke run would still "pass" by only printing.
+        import pyarrow.parquet as pq  # noqa: PLC0415
+
+        events = [
+            r
+            for shard in summary.written_paths
+            for ev in pq.read_table(shard).column("language_events").to_pylist()
+            for r in ev
+        ]
+        n_interjections = sum(1 for r in events if r.get("style") == "interjection")
+        n_speech = sum(1 for r in events if r.get("style") is None and r.get("role") == "assistant")
+        print(f"interjections={n_interjections} speech_atoms={n_speech}")
+        assert n_interjections > 0, "no interjection rows produced — check the interjection prompt marker"
+        assert n_speech > 0, "no speech tool-call atoms produced — check the speech prompt marker"
     return 0