fix: disable reward normalization for SWE configs with group_size=1

With group_size=1, mean normalization over traj_group_id produces
all-zero advantages (R - R = 0). Use method: identity to skip
normalization and pass raw rewards through directly.

Closes #397

Author: shamanez

examples/agentic_demo/agent_val_rock_swe.yaml

@@ -127,8 +127,8 @@ reference:
   infer_batch_size: 1
 
 reward_normalization:
-  grouping: traj_group_id # 可以tags(env_type)/traj_group_id(group)/batch(rollout_batch)... group_by计算reward/adv
-  method: mean
+  grouping: traj_group_id
+  method: identity
   # norm_mean_type: batch
   # norm_std_type: group
 

examples/agentic_demo/agent_val_rock_swe_qwen35_2b.yaml

@@ -138,7 +138,7 @@ reference:
 
 reward_normalization:
   grouping: traj_group_id
-  method: mean
+  method: identity
 
 train_env_manager:
   max_env_num_per_worker: 1