fix(hyperliquid): enforce 55s timeout for entire metric collection

- Wrap HTTP and WebSocket metric collection in asyncio.wait_for
- Remove redundant inner timeout configurations in aiohttp
- Reduce MAX_RETRIES from 3 to 2 and Retry-After from 15s to 3s
- Ensure total execution cannot exceed 55s including retries/sleeps

This prevents Hyperliquid metrics from exceeding the 59s Vercel limit
by enforcing the timeout at the outer scope rather than per-request.

🤖 Generated with [Claude Code](https://claude.ai/code)

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

Author: smypmsa

common/metric_types.py

@@ -13,7 +13,7 @@
 from common.metric_config import MetricConfig, MetricLabelKey, MetricLabels
 from common.metrics_handler import MetricsHandler
 
-MAX_RETRIES = 3
+MAX_RETRIES = 2
 
 
 class WebSocketMetric(BaseMetric):
@@ -59,18 +59,29 @@ async def collect_metric(self) -> None:
         """Collects single WebSocket message."""
         websocket = None
 
-        try:
+        async def _collect_ws_data():
+            nonlocal websocket
             websocket = await self.connect()
             await self.subscribe(websocket)
             data = await self.listen_for_data(websocket)
-
+            
             if data is not None:
-                latency: int | float = self.process_data(data)
-                self.update_metric_value(latency)
-                self.mark_success()
-                return
+                return data
             raise ValueError("No data in response")
 
+        try:
+            data = await asyncio.wait_for(
+                _collect_ws_data(),
+                timeout=self.config.timeout
+            )
+            latency: int | float = self.process_data(data)
+            self.update_metric_value(latency)
+            self.mark_success()
+
+        except asyncio.TimeoutError:
+            self.mark_failure()
+            self.handle_error(TimeoutError(f"WebSocket metric collection exceeded {self.config.timeout}s timeout"))
+
         except Exception as e:
             self.mark_failure()
             self.handle_error(e)
@@ -97,13 +108,19 @@ def get_endpoint(self) -> str:
 
     async def collect_metric(self) -> None:
         try:
-            data = await self.fetch_data()
+            data = await asyncio.wait_for(
+                self.fetch_data(),
+                timeout=self.config.timeout
+            )
             if data is not None:
                 latency: int | float = self.process_data(data)
                 self.update_metric_value(latency)
                 self.mark_success()
                 return
             raise ValueError("No data in response")
+        except asyncio.TimeoutError:
+            self.mark_failure()
+            self.handle_error(TimeoutError(f"Metric collection exceeded {self.config.timeout}s timeout"))
         except Exception as e:
             self.mark_failure()
             self.handle_error(e)
@@ -205,7 +222,6 @@ async def on_request_end(session, context, params):
         trace_config.on_request_end.append(on_request_end)
 
         async with aiohttp.ClientSession(
-            timeout=aiohttp.ClientTimeout(total=self.config.timeout),
             trace_configs=[trace_config],
         ) as session:
             response_time = 0.0
@@ -219,7 +235,7 @@ async def on_request_end(session, context, params):
                 response_time: float = time.monotonic() - start_time
 
                 if response.status == 429 and retry_count < MAX_RETRIES - 1:
-                    wait_time = int(response.headers.get("Retry-After", 15))
+                    wait_time = int(response.headers.get("Retry-After", 3))
                     await response.release()
                     await asyncio.sleep(wait_time)
                     continue
@@ -279,7 +295,6 @@ async def _send_request(
                 "Content-Type": "application/json",
             },
             json=self._base_request,
-            timeout=self.config.timeout,  # type: ignore
         )
 
     def process_data(self, value: float) -> float: