Adjust type hint of Norm.__call__ to return masked array

xref pydata/xarray#8030

We were perhaps overly permissive in the return type, I'm pretty sure we always return Masked Array, so may as well say so, so that downstream can rely on that (which they already are, to be clear...)

lib/matplotlib/colors.pyi

@@ -146,7 +146,7 @@ class Normalize:
     def clip(self, value: bool) -> None: ...
     @staticmethod
     def process_value(value: ArrayLike) -> tuple[np.ma.MaskedArray, bool]: ...
-    def __call__(self, value: ArrayLike, clip: bool | None = ...) -> ArrayLike: ...
+    def __call__(self, value: ArrayLike, clip: bool | None = ...) -> np.ma.MaskedArray: ...
     def inverse(self, value: ArrayLike) -> ArrayLike: ...
     def autoscale(self, A: ArrayLike) -> None: ...
     def autoscale_None(self, A: ArrayLike) -> None: ...
@@ -161,7 +161,7 @@ class TwoSlopeNorm(Normalize):
     @vcenter.setter
     def vcenter(self, value: float) -> None: ...
     def autoscale_None(self, A: ArrayLike) -> None: ...
-    def __call__(self, value: ArrayLike, clip: bool | None = ...) -> ArrayLike: ...
+    def __call__(self, value: ArrayLike, clip: bool | None = ...) -> np.ma.MaskedArray: ...
     def inverse(self, value: ArrayLike) -> ArrayLike: ...
 
 class CenteredNorm(Normalize):
@@ -240,7 +240,7 @@ class PowerNorm(Normalize):
         vmax: float | None = ...,
         clip: bool = ...,
     ) -> None: ...
-    def __call__(self, value: ArrayLike, clip: bool | None = ...) -> ArrayLike: ...
+    def __call__(self, value: ArrayLike, clip: bool | None = ...) -> np.ma.MaskedArray: ...
     def inverse(self, value: ArrayLike) -> ArrayLike: ...
 
 class BoundaryNorm(Normalize):
@@ -256,11 +256,11 @@ class BoundaryNorm(Normalize):
         *,
         extend: Literal["neither", "both", "min", "max"] = ...
     ) -> None: ...
-    def __call__(self, value: ArrayLike, clip: bool | None = ...) -> ArrayLike: ...
+    def __call__(self, value: ArrayLike, clip: bool | None = ...) -> np.ma.MaskedArray: ...
     def inverse(self, value: ArrayLike) -> ArrayLike: ...
 
 class NoNorm(Normalize):
-    def __call__(self, value: ArrayLike, clip: bool | None = ...) -> ArrayLike: ...
+    def __call__(self, value: ArrayLike, clip: bool | None = ...) -> np.ma.MaskedArray: ...
     def inverse(self, value: ArrayLike) -> ArrayLike: ...
 
 def rgb_to_hsv(arr: ArrayLike) -> np.ndarray: ...