Check list multiplication operands in evaluation order (#22263)

Check the list first, using the outer context because list repetition
preserves the item type (this code path doesn't look at rmul), and let
the operator checker handle the multiplier. Also the code path works for
SupportsIndex now

Authored by Codex

Author: hauntsaninja

mypy/checkexpr.py

@@ -4537,13 +4537,7 @@ def check_list_multiply(self, e: OpExpr) -> Type:
 
         Type inference is special-cased for this common construct.
         """
-        right_type = self.accept(e.right)
-        if is_subtype(right_type, self.named_type("builtins.int")):
-            # Special case: [...] * <int value>. Use the type context of the
-            # OpExpr, since the multiplication does not affect the type.
-            left_type = self.accept(e.left, type_context=self.type_context[-1])
-        else:
-            left_type = self.accept(e.left)
+        left_type = self.accept(e.left, type_context=self.type_context[-1])
         result, method_type = self.check_op("__mul__", left_type, e.right, e)
         e.method_type = method_type
         return result

test-data/unit/check-inference-context.test

@@ -847,6 +847,12 @@ a: List[int]
 if int():
     a = [None] * 3  # E: List item 0 has incompatible type "None"; expected "int"
     a = [''] * 3  # E: List item 0 has incompatible type "str"; expected "int"
+
+def walrus(n: int | None) -> None:
+    a: List[int] = [n] * (n := 1)  # E: List item 0 has incompatible type "int | None"; expected "int"
+    n = None
+    b: List[object] = [(n := 1)] * n
+    reveal_type(n)  # N: Revealed type is "builtins.int"
 [builtins fixtures/list.pyi]
 
 [case testUnionTypeContext]