hdl.ast: Test *Castable subclasses on definition.

The __init_subclass__ method fires on class definition rather than use.
It also has the bonus impact that no __new__ method is defined, so the
classes can be correctly detected as mix-in classes by modules such as
enum.

amaranth/hdl/ast.py

@@ -41,12 +41,10 @@ class ShapeCastable:
     a richer description of the shape than what is supported by the core Amaranth language, yet
     still be transparently used with it.
     """
-    def __new__(cls, *args, **kwargs):
-        self = super().__new__(cls)
-        if not hasattr(self, "as_shape"):
+    def __init_subclass__(cls, **kwargs):
+        if not hasattr(cls, "as_shape"):
             raise TypeError(f"Class '{cls.__name__}' deriving from `ShapeCastable` must override "
                             f"the `as_shape` method")
-        return self
 
 
 class Shape:
@@ -1319,15 +1317,13 @@ class ValueCastable:
     from :class:`ValueCastable` is mutable, it is up to the user to ensure that it is not mutated
     in a way that changes its representation after the first call to :meth:`as_value`.
     """
-    def __new__(cls, *args, **kwargs):
-        self = super().__new__(cls)
-        if not hasattr(self, "as_value"):
+    def __init_subclass__(cls, **kwargs):
+        if not hasattr(cls, "as_value"):
             raise TypeError(f"Class '{cls.__name__}' deriving from `ValueCastable` must override "
                             "the `as_value` method")
-        if not hasattr(self.as_value, "_ValueCastable__memoized"):
+        if not hasattr(cls.as_value, "_ValueCastable__memoized"):
             raise TypeError(f"Class '{cls.__name__}' deriving from `ValueCastable` must decorate "
                             "the `as_value` method with the `ValueCastable.lowermethod` decorator")
-        return self
 
     @staticmethod
     def lowermethod(func):