only benchmark relevant tests to get rid of noise

Author: mohammedahmed18

codeflash/api/aiservice.py

@@ -49,7 +49,7 @@ def __init__(self) -> None:
         self.is_local = self.base_url == "http://localhost:8000"
         # (connect_timeout, read_timeout) — connect should be fast; read
         # can be slow because the server runs LLM inference.
-        self.timeout: float | tuple[float, float] | None = (10, 600)
+        self.timeout: float | tuple[float, float] | None = (10, 300)
 
     def get_next_sequence(self) -> int:
         """Get the next LLM call sequence number."""

codeflash/cli_cmds/cli.py

@@ -124,6 +124,7 @@ def process_pyproject_config(args: Namespace) -> Namespace:
 
     if args.tests_root is None:
         if is_go_project:
+            # this is just a placeholder, in go we put generated test files in the same package as the source
             args.tests_root = args.module_root
         elif is_java_project:
             # Try standard Maven/Gradle test directories