feat(V2): /health/ready con ping client + middleware request log strutturato + request_id correlation

- /health/ready: ping di tutti i client (exchange, env) cached con
  timeout 2s, status ready|degraded|not_ready, opt-in 503 via
  READY_FAILS_ON_DEGRADED.
- Middleware mcp.request: 1 riga JSON per HTTP request con request_id,
  method, path, status_code, duration_ms, actor, bot_tag, exchange,
  tool, client_ip, user_agent.
- request_id propagato in request.state, audit log e error envelope per
  correlazione cross-cutting.
- Aggiunto async health() come probe minimo a bybit/alpaca/macro/
  sentiment/deribit (hyperliquid lo aveva già).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/cerbero_mcp/server.py

@@ -1,7 +1,9 @@
 """Factory FastAPI app con middleware, swagger, exception handlers."""
 from __future__ import annotations
 
+import asyncio
 import json
+import os
 import time
 from datetime import UTC, datetime
 from typing import Any
@@ -18,6 +20,7 @@ from cerbero_mcp.common.errors import (
     RETRYABLE_STATUSES,
     error_envelope,
 )
+from cerbero_mcp.common.request_log import install_request_log_middleware
 
 
 class _TimestampInjectorMiddleware(BaseHTTPMiddleware):
@@ -99,6 +102,11 @@ def build_app(
         app, testnet_token=testnet_token, mainnet_token=mainnet_token
     )
 
+    # Request log middleware: registrato DOPO auth → starlette esegue
+    # i middleware in ordine inverso (LIFO) → request_log è outermost,
+    # auth è interno e popola request.state.* prima del ritorno.
+    install_request_log_middleware(app)
+
     app.add_middleware(_TimestampInjectorMiddleware)
 
     @app.middleware("http")
@@ -128,6 +136,7 @@ def build_app(
             content=error_envelope(
                 type_="http_error", code=code, message=message,
                 retryable=retryable, details=details,
+                request_id=getattr(request.state, "request_id", None),
             ),
         )
 
@@ -155,6 +164,7 @@ def build_app(
                 message=f"request body validation failed on {first_loc}",
                 retryable=False, suggested_fix=suggestion,
                 details={"errors": safe_errs},
+                request_id=getattr(request.state, "request_id", None),
             ),
         )
 
@@ -166,6 +176,7 @@ def build_app(
                 type_="internal_error", code="UNHANDLED_EXCEPTION",
                 message=f"{type(exc).__name__}: {str(exc)[:300]}",
                 retryable=True,
+                request_id=getattr(request.state, "request_id", None),
             ),
         )
 
@@ -179,6 +190,79 @@ def build_app(
             "data_timestamp": datetime.now(UTC).isoformat(),
         }
 
+    @app.get("/health/ready", tags=["system"])
+    async def health_ready():
+        """Readiness check: ping ogni client exchange cached.
+
+        - Itera ``app.state.registry._clients`` (se presente).
+        - Per ogni client prova ``health()`` (preferito) o ``is_testnet()``.
+          In assenza di metodo, marca con ``note: no probe method``.
+        - Timeout di 2s per client tramite ``asyncio.wait_for``.
+        - Stato globale: ``ready`` se tutti ok, ``degraded`` se almeno
+          uno fallisce, ``not_ready`` se registry vuoto.
+        - HTTP 200 di default; con ``READY_FAILS_ON_DEGRADED=true`` ritorna
+          503 quando lo stato non è ``ready`` (utile per probe k8s).
+        """
+        registry = getattr(app.state, "registry", None)
+        clients_status: list[dict[str, Any]] = []
+        if registry is not None:
+            for (exchange, env), client in registry._clients.items():
+                t0 = time.perf_counter()
+                ping = (
+                    getattr(client, "health", None)
+                    or getattr(client, "is_testnet", None)
+                )
+                if ping is None:
+                    clients_status.append({
+                        "exchange": exchange,
+                        "env": env,
+                        "healthy": True,
+                        "note": "no probe method",
+                    })
+                    continue
+                try:
+                    res = ping()
+                    if asyncio.iscoroutine(res):
+                        await asyncio.wait_for(res, timeout=2.0)
+                    dur = (time.perf_counter() - t0) * 1000
+                    clients_status.append({
+                        "exchange": exchange,
+                        "env": env,
+                        "healthy": True,
+                        "duration_ms": round(dur, 2),
+                    })
+                except Exception as e:
+                    clients_status.append({
+                        "exchange": exchange,
+                        "env": env,
+                        "healthy": False,
+                        "error": f"{type(e).__name__}: {str(e)[:200]}",
+                    })
+
+        if not clients_status:
+            status_label = "not_ready"
+        elif all(c["healthy"] for c in clients_status):
+            status_label = "ready"
+        else:
+            status_label = "degraded"
+
+        fail_on_degraded = os.environ.get(
+            "READY_FAILS_ON_DEGRADED", "false"
+        ).lower() in ("1", "true", "yes")
+        http_code = 200
+        if fail_on_degraded and status_label != "ready":
+            http_code = 503
+
+        body = {
+            "status": status_label,
+            "name": title,
+            "version": version,
+            "uptime_seconds": int(time.time() - app.state.boot_at),
+            "data_timestamp": datetime.now(UTC).isoformat(),
+            "clients": clients_status,
+        }
+        return JSONResponse(status_code=http_code, content=body)
+
     def _custom_openapi() -> dict[str, Any]:
         if app.openapi_schema:
             return app.openapi_schema