feat: Primary-Wait backoff queuing — BIZ-60

When all primary backends are in cooldown, wait and retry the primary pool
before falling through to fallback/emergency. This reduces unnecessary
spend on paid fallback providers during temporary 429 storms.

Config:
- primary_wait_ms (default 5000, env SIDECAR_PRIMARY_WAIT_MS)
- primary_wait_max_retries (default 6, env SIDECAR_PRIMARY_WAIT_MAX_RETRIES)

Implementation:
- config.py: 2 new config fields + env var loading
- router.py: pick_primary_backend() — primary-pool-only selection
- proxy.py: primary-wait loop between standard retries and emergency

Expected win: 17% error rate during high concurrency drops, emergency
passthrough count falls as requests wait for NVIDIA pool recovery
instead of immediately routing to SiliconFlow fallback.

config.py

@@ -73,6 +73,10 @@ class Config:
     # Stats
     stats_refresh_interval_seconds: float = 30.0
 
+    # Primary-Wait: when all primary backends are cooling, wait before fallback
+    primary_wait_ms: int = 5000
+    primary_wait_max_retries: int = 6
+
     # Request timeout
     default_request_timeout_seconds: int = 120
 
@@ -114,6 +118,16 @@ class Config:
         # Logging
         c.log_level = os.getenv("LOG_LEVEL", c.log_level).upper()
 
+        # Primary-Wait
+        c.primary_wait_ms = int(
+            os.getenv("SIDECAR_PRIMARY_WAIT_MS", str(c.primary_wait_ms))
+        )
+        c.primary_wait_max_retries = int(
+            os.getenv(
+                "SIDECAR_PRIMARY_WAIT_MAX_RETRIES", str(c.primary_wait_max_retries)
+            )
+        )
+
         # Database
         c.db_path = os.getenv(
             "SIDECAR_DB_PATH",