svcforge: reference implementation

Complete working build of the system learn-python/ teaches.
164 tests, mypy --strict clean, domain coverage 99%.

migrations/001_init.sql

@@ -0,0 +1,36 @@
+-- 001_init.sql — instances and tasks.
+--
+-- Forward-only. There is no down script. If this is wrong, 002 fixes it.
+
+create table instances (
+  id           uuid primary key default gen_random_uuid(),
+  team         text        not null,
+  service_type text        not null,           -- 'elasticsearch' | 'redis' | ...
+  size         text        not null,           -- 'small' | 'medium'
+  state        text        not null,
+  namespace    text        not null,
+  release_name text        not null unique,    -- helm release; the idempotency anchor
+  chart_version text       not null,           -- what's ACTUALLY deployed. day-2 hinges on this column.
+  endpoint     text,
+  error        text,
+  expires_at   timestamptz,
+  created_at   timestamptz not null default now(),
+  updated_at   timestamptz not null default now()
+);
+
+create table tasks (
+  id            bigserial primary key,
+  instance_id   uuid        not null references instances(id) on delete cascade,
+  kind          text        not null,          -- 'provision'|'deprovision'|'upgrade'|'verify'
+  state         text        not null default 'queued',  -- queued|running|done|failed
+  attempts      int         not null default 0,
+  run_after     timestamptz not null default now(),     -- backoff lands here
+  locked_by     text,
+  locked_at     timestamptz,
+  last_error    text,
+  created_at    timestamptz not null default now()
+);
+
+-- Partial index: the claim query only ever looks at queued rows. Keeping the index
+-- to that subset means it stays small no matter how much history `tasks` accumulates.
+create index tasks_runnable on tasks (run_after) where state = 'queued';