feat(agent-call): add contract-driven mock executor

docs/contracts/examples/README.md

@@ -0,0 +1,5 @@
+# Contract fixtures
+
+`call.execute.json` is the canonical valid command fixture. Runtime tests generate the remaining event fixtures from persisted facts so replay assertions compare original bytes and IDs rather than synthesized history. Invalid cases include unknown schema versions, missing required fields, cross-tenant bindings, conflicting idempotency bodies, and tenant routing keys over 224 UTF-8 bytes.
+
+All fixtures are synthetic. The profile is `mock`; it is never a production provider configuration.

docs/contracts/examples/call.execute.json

@@ -0,0 +1,23 @@
+{
+  "schema_version": "1.0",
+  "command_type": "call.execute",
+  "command_id": "cmd_demo_001",
+  "tenant_id": "tenant-demo",
+  "tenant_key": "tenant-demo-key",
+  "trace_id": "trace_demo_001",
+  "issued_at": "2026-09-11T08:00:00Z",
+  "not_after": "2099-09-11T08:05:00Z",
+  "payload": {
+    "execution_id": "exec_demo_001",
+    "task_id": "task-demo",
+    "task_item_id": "item_demo",
+    "task_revision": 1,
+    "callee": "18601013734",
+    "route_policy_id": "route_policy_test",
+    "caller_profile_id": "caller_profile_test",
+    "agent_version_id": "agent_v1",
+    "variables": {},
+    "ring_timeout_ms": 30000,
+    "max_call_duration_ms": 180000
+  }
+}