Add automatic data-attribute-to-Postgres sync

[longquanzheng]

What

Adds an opt-in, SDK-side feature to automatically mirror a workflow data attribute to a single cell (row + column) of a user-owned Postgres table:

• Load on start — the data attribute is loaded from its mapped column when the workflow starts.
• Sync on mutation — when the data attribute is mutated in a state's execute, the new value is written back to the column.

The iWF data attribute stays the source of truth; the DB cell is a mirror. The SDK is driver-agnostic (depends only on javax.sql.DataSource); the user owns the schema, credentials, and pooling.

API

DataAttributeDef.create(String.class, "status",
    DbAttributeSync.of(dataSource, "public.orders", "id",
        (workflowId, persistence) -> new CellLocation(workflowId, "status_col")));

How it works

• On start, Client swaps the start state to a worker-internal load state that SELECTs each mapped column, sets the data attribute, then transitions to the real starting state (original input passed through).
• When a mapped data attribute is mutated in execute, the worker reroutes the decision through a worker-internal sync state (carrying the captured original decision as DbSyncStatePayload) that UPDATEs the column, then replays the original decision verbatim — so any decision shape (single/multi-next, graceful/force complete/fail, conditionalClose) is preserved.
• iWF persists the data attribute first, then the DB write happens, so a DB failure retries only the sync step and never re-runs user execute code.
• Neither system state is registered in the Registry; both are handled specially in WorkerService by reserved stateId.

Security

• Credentials never held by the SDK (user-supplied DataSource).
• All values bound via PreparedStatement; identifiers validated against an allowlist and double-quoted.

Tests / CI

• DbSyncTest runs end-to-end against a real Postgres (script/docker-compose.yml iwf-integ-postgres + script/postgres-init/iwf-integ-init.sql) and a live iWF server; verified locally (load + sync).
• CI adds a Postgres readiness wait before the build.

Scope (v1)

• Fixed-key data attributes only (not createByPrefix).
• Sync on execute mutations only (not waitUntil).
• Load fires for starts via the registered Client.
• The mapped column stores the attribute's JSON — use text/jsonb.

Docs

• User guide and overview, linked from the README.