[torch.compile] remove compilation_context and simplify code (#10838)

Signed-off-by: youkaichao <youkaichao@gmail.com>

vllm/compilation/backends.py

@@ -242,10 +242,6 @@ class VllmBackend:
         assert not self._called, "VllmBackend can only be called once"
 
         self.graph = graph
-        # config is updated now, because only here can
-        # we get the sizes to capture for cudagraph
-        # from compilation context
-        self.compilation_configs.init_during_runtime()
         self.configure_post_pass()
 
         self.split_gm, self.piecewise_graphs = split_graph(

vllm/compilation/compile_context.py

@@ -1,23 +0,0 @@
-from contextlib import contextmanager
-from typing import Any
-
-_compile_context: Any = None
-
-
-def get_compile_context() -> Any:
-    """Get the current compile context."""
-    return _compile_context
-
-
-@contextmanager
-def set_compile_context(context: Any):
-    """A context manager that stores the current compile context,
-    usually it is a list of sizes to specialize.
-    """
-    global _compile_context
-    prev_context = _compile_context
-    _compile_context = context
-    try:
-        yield
-    finally:
-        _compile_context = prev_context