feat(db): discovery_snapshots table — daily Discovery-Tracking storage (P2 sub-step)

[MoltyCel]

Summary

P2 Sub-Step (a) per Discovery-Tracking-Baseline SPEC (PR #54 merged): neue Tabelle discovery_snapshots für tägliche Discovery-Surface-Snapshots.

Klein + fokussiert — nur CREATE TABLE + INDEX + COMMENTs. Kein Code-Change. Idempotent (IF NOT EXISTS).

Schema

CREATE TABLE discovery_snapshots (
  id           BIGSERIAL PRIMARY KEY,
  snapshot_at  DATE        NOT NULL UNIQUE,        -- one row per day
  generated_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
  payload      JSONB       NOT NULL,
  source_run_status TEXT   NOT NULL DEFAULT 'ok'
    CHECK (source_run_status IN ('ok','partial','failed'))
);

CREATE INDEX idx_discovery_snapshots_at ON discovery_snapshots(snapshot_at DESC);

JSONB-payload per SPEC §3.5 (5 Sub-Sections: self_probes, gsc, bot_hits, github, errors).

Pre-Commit-Diff (§8)

 migrations/2026-05-21_create_discovery_snapshots.sql | 30 ++++++++++++++++++++++
 1 file changed, 30 insertions(+)

Genau 1 File, etabliertes migrations/YYYY-MM-DD_<desc>.sql-Schema, kein Fremd-Scope.

Branch-Hygiene (§11.4)

Branch ab frischem origin/main (cd1b0e5, 0 behind), Worktree ~/moltrust-api-I.

§2.3 Cross-Review

Skip — reine Schema-Definition, kein Auth-/Credential-/Token-Pfad. Tabelle ist Container für eigene aggregierte Metriken.

Test plan

• Merge per Merge-Commit
• Apply via psql -d moltstack -f migrations/2026-05-21_create_discovery_snapshots.sql
• Verify: \d discovery_snapshots zeigt 5 Spalten + Index + Comments
• Folge-Sub-Step: Baseline-Snapshot manuell einsammeln + INSERTen (one-off, kein Repo-Track)