[ROCm][CI] Fix entrypoints tests and Python-only installation test on ROCm (#28979)

Signed-off-by: Andreas Karatzas <akaratza@amd.com>

tests/entrypoints/pooling/basic/test_truncation.py

@@ -9,11 +9,6 @@ import pytest_asyncio
 from tests.utils import RemoteOpenAIServer
 from vllm.platforms import current_platform
 
-if current_platform.is_rocm():
-    pytest.skip(
-        "Encoder self-attention is not implemented on ROCm.", allow_module_level=True
-    )
-
 MODEL_NAME = "sentence-transformers/all-MiniLM-L12-v2"
 max_model_len = 128
 
@@ -44,6 +39,10 @@ def server():
         str(max_model_len),
     ]
 
+    # ROCm: Use Flex Attention to support encoder-only self-attention.
+    if current_platform.is_rocm():
+        args.extend(["--attention-backend", "FLEX_ATTENTION"])
+
     with RemoteOpenAIServer(MODEL_NAME, args) as remote_server:
         yield remote_server