[Bugfix][TPU] Fix megacore setting for v5e-litepod (vllm-project#6397)

vllm/attention/backends/pallas.py

@@ -116,7 +116,7 @@ def __init__(
 
         self.megacore_mode = None
         tpu_type = torch_xla.tpu.get_tpu_env()["TYPE"].lower()
-        if not tpu_type.endswith("lite"):
+        if "lite" not in tpu_type:
             if self.num_kv_heads % 2 == 0:
                 self.megacore_mode = "kv_head"
             else: