feat(snippets): add notes.data JSONB — the indexed mirror of snippet fields

Milestone #232 step 1 (task #2081). Takes the enabler first rather than the
write-path trigger: reverse lookup, drift checks and the duplicate finder all
need to QUERY structured fields, and building them on body-regex first means
writing them twice.

#227 deferred this bag "unless body-convention ergonomics prove insufficient" —
answering "which snippets live in this file?" by scanning every snippet and
regexing its body is that condition being met.

Migration 0070 adds `notes.data` (nullable JSONB) + a GIN index. The body is
UNCHANGED and still what gets embedded and read by humans; `data` mirrors the
same facts in a shape Postgres can index. Code is deliberately not copied into
it — the body holds it, and duplicating a blob into the column we index around
would be waste.

- compose_data() builds the mirror, omitting empties so the column stays sparse
- snippet_fields() prefers `data`, falling back to parsing the body. Rows written
  before 0070 have no `data` and are never backfilled, so a hand-edited body
  stays authoritative for them with no conversion deadline
- create / update / merge all write body and mirror from the same merged field
  set, so the two can't drift; merge in particular has to grow the mirror with
  the survivor's location set or a merged snippet would be unfindable at the very
  call sites the merge just recorded

Named `data`, not `metadata`, because that collides with SQLAlchemy's declarative
Base.metadata — which is why the pre-0069 model had to map an awkward
`entity_metadata` attribute. Not a revival of the column 0069 dropped: different
name, different purpose, nothing reads the old shape.

Two test fakes needed an explicit `data = None`: snippet_fields prefers `data`
when truthy and an auto-MagicMock attribute is truthy, so every parsed field
would have come back a MagicMock. Checked every fake reaching snippet code this
time rather than waiting for CI (note 2109).

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RLwAaV4DQEmVyn496HnEvt

alembic/versions/0070_notes_data_jsonb.py

@@ -0,0 +1,60 @@
+"""add notes.data JSONB — queryable structured fields for typed records
+
+Revision ID: 0070
+Revises: 0069
+Create Date: 2026-07-26
+
+Snippets (note_type='snippet') carry structured fields — name, language,
+signature, and a list of canonical locations (repo · path · symbol). Those were
+stored as a markdown body-convention, which reads well and feeds the embedding
+but cannot be QUERIED: answering "which snippets live in this file?" meant
+scanning every snippet and regexing its body.
+
+This adds a general `data` JSONB column plus a GIN index, so those fields become
+indexable. The body stays exactly as it was — it is still the human-readable
+form and still what gets embedded. `data` is a queryable mirror of the same
+facts, not a replacement, and the code itself is deliberately NOT copied into it
+(the body already holds it; duplicating a blob to index fields around it would
+be waste).
+
+Relationship to 0069: that migration DROPPED `notes.metadata`, a JSONB column
+which only ever held person/place/list entity fields, when those surfaces were
+removed. This is not a revival of it — different name, different purpose, and
+nothing reads the old shape. The column is named `data` rather than `metadata`
+because `metadata` collides with SQLAlchemy's declarative `Base.metadata`, which
+is why the old model had to map an awkward `entity_metadata` attribute onto it.
+
+Nullable with no backfill, deliberately: rows written before this migration keep
+working because the service falls back to parsing the body when `data` is
+absent. That means no migration deadline and no risk of a backfill mangling a
+hand-edited body.
+
+Downgrade drops the index and the column. Any structured fields it held remain
+recoverable from the body convention, which is the same source they mirror.
+"""
+from alembic import op
+import sqlalchemy as sa
+from sqlalchemy.dialects.postgresql import JSONB
+
+
+revision = "0070"
+down_revision = "0069"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.add_column("notes", sa.Column("data", JSONB, nullable=True))
+    # GIN supports containment (`data @> '{"locations":[{"repo":"x"}]}'`), which
+    # covers exact repo/path/symbol and language lookups. Path PREFIX matching
+    # ("everything under frontend/src/") is not an index-served operation here
+    # and still filters after the fact — acceptable while snippet counts are
+    # small, and a generated column is the escape hatch if that changes.
+    op.create_index(
+        "ix_notes_data_gin", "notes", ["data"], postgresql_using="gin",
+    )
+
+
+def downgrade() -> None:
+    op.drop_index("ix_notes_data_gin", table_name="notes")
+    op.drop_column("notes", "data")