Avoid GCC -Warray-bounds false positive on fixed-size shape copies

include/xtensor/core/xfunction.hpp

@@ -712,7 +712,7 @@ namespace xt
     {
         if (m_cache.is_initialized && reuse_cache)
         {
-            std::copy(m_cache.shape.cbegin(), m_cache.shape.cend(), shape.begin());
+            std::copy_n(m_cache.shape.cbegin(), m_cache.shape.size(), shape.begin());
             return m_cache.is_trivial;
         }
         else

include/xtensor/core/xmath.hpp

@@ -1960,7 +1960,7 @@ namespace xt
                 );
             }
 
-            std::copy(e.shape().begin(), e.shape().end(), broadcast_shape.begin());
+            std::copy_n(e.shape().begin(), e.shape().size(), broadcast_shape.begin());
         }
 
         constexpr layout_type L = default_assignable_layout(std::decay_t<W>::static_layout);

include/xtensor/misc/xmanipulation.hpp

@@ -840,12 +840,12 @@ namespace xt
 
         shape_type shape;
         resize_container(shape, e.shape().size());
-        std::copy(e.shape().cbegin(), e.shape().cend(), shape.begin());
+        std::copy_n(e.shape().cbegin(), e.shape().size(), shape.begin());
 
         get_strides_t<shape_type> strides;
         decltype(auto) old_strides = detail::get_strides<XTENSOR_DEFAULT_LAYOUT>(e);
         resize_container(strides, old_strides.size());
-        std::copy(old_strides.cbegin(), old_strides.cend(), strides.begin());
+        std::copy_n(old_strides.cbegin(), old_strides.size(), strides.begin());
 
         strides[axis] *= -1;
         std::size_t offset = static_cast<std::size_t>(

include/xtensor/views/xbroadcast.hpp

@@ -293,7 +293,7 @@ namespace xt
             XTENSOR_THROW(xt::broadcast_error, "Broadcast shape has fewer elements than original expression.");
         }
         xt::resize_container(m_shape, s.size());
-        std::copy(s.begin(), s.end(), m_shape.begin());
+        std::copy_n(s.begin(), s.size(), m_shape.begin());
         xt::broadcast_shape(m_e.shape(), m_shape);
     }