P0 security batch: admin-only user mgmt, unified role model, login rate limiting, webhook secret, security headers, pagination caps

- Remove POST /api/auth/register (404); no sign-up UI; users are admin-managed
- Add admin-only POST/PATCH/DELETE /api/auth/users (forced canonical roles,
  self-lockout + reference guards)
- Unify role model in app/core/roles.py; reject unknown roles at creation and
  at login/JWT validation; startup normalizes unambiguous legacy aliases
- Login rate limiting ~5 fails/15 min per IP+email -> 429 (in-process, tunable)
- WhatsApp webhook requires X-Webhook-Secret; fail-closed when env unset;
  GET handshake uses constant-time verify token (403 on mismatch)
- GET /api/whatsapp/mock-log now requires auth
- Security headers middleware: X-Frame-Options DENY, nosniff, CSP on HTML,
  HSTS behind TLS
- Pagination: limit alias for page_size, hard cap enforced, both -> 422

app/core/ratelimit.py

@@ -0,0 +1,82 @@
+"""Dependency-light login rate limiter.
+
+Brute-force protection for ``POST /api/auth/login``: a sliding window of
+failed attempts keyed by ``ip|email``.  Defaults to ~5 failures / 15 minutes
+(env-tunable via ``LOGIN_RATE_LIMIT_MAX_ATTEMPTS`` /
+``LOGIN_RATE_LIMIT_WINDOW_SECONDS``).
+
+In-process storage is intentional: the app currently runs a single uvicorn
+worker, and keeping the limiter dependency-light avoids pulling slowapi in
+for one endpoint.  ``_now`` is a module-level hook so tests can fast-forward
+the clock.
+"""
+
+from __future__ import annotations
+
+import time
+from collections import defaultdict, deque
+
+from fastapi import HTTPException, status
+
+from app.core.config import settings
+
+
+def _now() -> float:
+    """Wall-clock epoch seconds; overridable in tests via monkeypatch."""
+    return time.time()
+
+
+class LoginRateLimiter:
+    """Sliding-window failure limiter keyed by ``ip|email``."""
+
+    def __init__(self, max_attempts: int = 5, window_seconds: int = 15 * 60) -> None:
+        self.max_attempts = max(max_attempts, 1)
+        self.window_seconds = max(window_seconds, 1)
+        self._failures: defaultdict[str, deque[float]] = defaultdict(deque)
+
+    def key(self, ip: str, email: str) -> str:
+        return f"{ip}|{email.strip().lower()}"
+
+    def _prune(self, key: str, now: float | None = None) -> None:
+        now = now if now is not None else _now()
+        window_start = now - self.window_seconds
+        bucket = self._failures.get(key)
+        if bucket is None:
+            return
+        while bucket and bucket[0] <= window_start:
+            bucket.popleft()
+        if not bucket:
+            self._failures.pop(key, None)
+
+    def failure_count(self, key: str) -> int:
+        self._prune(key)
+        return len(self._failures.get(key, ()))
+
+    def is_blocked(self, key: str) -> bool:
+        return self.failure_count(key) >= self.max_attempts
+
+    def record_failure(self, key: str) -> None:
+        self._failures[key].append(_now())
+        self._prune(key)
+
+    def clear(self, key: str) -> None:
+        self._failures.pop(key, None)
+
+    def reset(self) -> None:
+        self._failures.clear()
+
+    def check_or_raise(self, key: str) -> None:
+        """Raise HTTP 429 when the key has exhausted its attempts."""
+        if self.is_blocked(key):
+            raise HTTPException(
+                status_code=status.HTTP_429_TOO_MANY_REQUESTS,
+                detail="Too many failed login attempts. Try again later.",
+            )
+
+
+# Shared instance — module import is safe because the app fails closed at
+# boot (app/core/config.py) before any request can reach the login route.
+login_rate_limiter = LoginRateLimiter(
+    max_attempts=settings.LOGIN_RATE_LIMIT_MAX_ATTEMPTS,
+    window_seconds=settings.LOGIN_RATE_LIMIT_WINDOW_SECONDS,
+)