feat(agents): manifest plumbing + GET /agents/registry (ADR-0014 step 3)

Each agent now exports a module-level MANIFEST declaring id, version,
pref_schema, required_consents, ttl_sec, and silenced_in_contexts. The
registry surfaces both the agent and its manifest, and rejects on
mismatch so the two cannot drift.

ml/serving exposes GET /agents/registry; services/api proxies it as
GET /api/agents/registry with a 60s in-process cache so admin pageviews
don't hammer upstream. Failures aren't cached.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

ml/agents/momentum.py

@@ -1,13 +1,38 @@
 from __future__ import annotations
 from typing import ClassVar
 from .base import BaseAgent, AgentInput, AgentOutput
+from .manifest import AgentManifest
+
+
+MANIFEST = AgentManifest(
+    id="momentum",
+    version="1.0.0",
+    description="Characterises the user's recent engagement trend from profile features.",
+    pref_schema={
+        "type": "object",
+        "additionalProperties": False,
+        "properties": {
+            "low_engagement_threshold_pct": {
+                "type": "integer",
+                "minimum": 0,
+                "maximum": 100,
+                "default": 25,
+                "description": "Completion rate below which momentum hints at low engagement.",
+            },
+        },
+    },
+    context_schema=["profile.features"],
+    required_consents=["data:core", "agent:momentum"],
+    output_contract={"type": "snippet", "format": "free_text"},
+    ttl_sec=21_600,
+)
 
 
 class MomentumAgent(BaseAgent):
     """Characterises the user's recent engagement trend from profile features."""
-    agent_id: ClassVar[str] = "momentum"
-    ttl_seconds: ClassVar[int] = 21600  # 6h
-    version: ClassVar[str] = "1.0.0"
+    agent_id: ClassVar[str] = MANIFEST.id
+    ttl_seconds: ClassVar[int] = MANIFEST.ttl_sec
+    version: ClassVar[str] = MANIFEST.version
 
     def compute(self, inp: AgentInput) -> AgentOutput:
         completion = inp.profile.get("completion_rate_30d")