[Bugfix] tpu_model_runner: set vllm config context when calling reset_dynamo_cache() (#30331)

Signed-off-by: Daniele Trifirò <dtrifiro@redhat.com>

vllm/v1/worker/tpu_worker.py

@@ -10,7 +10,7 @@ import torch
 import torch.nn as nn
 
 import vllm.envs as envs
-from vllm.config import VllmConfig
+from vllm.config import VllmConfig, set_current_vllm_config
 from vllm.distributed import (
     ensure_model_parallel_initialized,
     init_distributed_environment,
@@ -207,6 +207,7 @@ class TPUWorker:
         # one compiled bytecode. Having one FX graph/cached bytecode per
         # compiled model is required for `support_torch_compile` decorator to
         # skip dynamo guard.
+        with set_current_vllm_config(self.vllm_config):
             self.model_runner.reset_dynamo_cache()
 
         # Get the maximum amount of memory used by the model weights and