fix(cache-middleware): wire max_entries through register.py

Per CodeRabbit on #1879: CacheMiddlewareConfig.max_entries was added
to the config schema but register.py still instantiated CacheMiddleware
with only (enabled_mode, similarity_threshold) — so a configured
max_entries value was silently ignored and deployments always used the
1024 default.

Pass config.max_entries through to the constructor so config-driven
LRU sizing actually takes effect.

Signed-off-by: ColinM-sys <cmcdonough@50words.com>

Author: ColinM-sys

packages/nvidia_nat_core/src/nat/middleware/cache/register.py

@@ -30,4 +30,8 @@ async def cache_middleware(config: CacheMiddlewareConfig, builder: Builder):
     Yields:
         A configured cache middleware instance
     """
-    yield CacheMiddleware(enabled_mode=config.enabled_mode, similarity_threshold=config.similarity_threshold)
+    yield CacheMiddleware(
+        enabled_mode=config.enabled_mode,
+        similarity_threshold=config.similarity_threshold,
+        max_entries=config.max_entries,
+    )