cutlass 3.9 update (#2255)

* cutlass 3.9 update

* rebase

* fixes out of shared memory for blockwise Blackwell

* doc format

* fix issue 2253

* disable host ref by default

* fix sm120 smem capacity

---------

Co-authored-by: yuzhai <yuzhai@nvidia.com>
Co-authored-by: Haicheng Wu <haichengw@nvidia.com>

include/cute/swizzle_layout.hpp

@@ -128,8 +128,6 @@ make_fragment_like(ComposedLayout<Swizzle<B,M,S>,Offset,Layout> const& layout)
 // Utilities
 //
 
-namespace detail {
-
 // Get just the Swizzle part of a composed layout.
 template <int B, int M, int S, class Offset, class LayoutB>
 CUTE_HOST_DEVICE constexpr
@@ -167,8 +165,6 @@ get_nonswizzle_portion(Layout<Shape,Stride> const& slayout)
   return slayout;
 }
 
-} // namespace detail
-
 //
 // Slice a Swizzled ComposedLayout
 //