optimization: schema: use slices.Sort instead of sort.Slice

Shaves off ~1/3rd of the time of Metadata.Equal for me; sort.Slice takes
in `any` as the type of the slice, which requires converting the slice to
an interface - which itself requires allocating and converting the
pointer to the slice header to use as the interface value.

Author: Noam Preil

arrow/schema.go

@@ -18,6 +18,7 @@ package arrow
 
 import (
 	"fmt"
+	"slices"
 	"sort"
 	"strings"
 
@@ -129,8 +130,8 @@ func (md Metadata) sortedIndices() []int {
 		idxes[i] = i
 	}
 
-	sort.Slice(idxes, func(i, j int) bool {
-		return md.keys[idxes[i]] < md.keys[idxes[j]]
+	slices.SortFunc(idxes, func(i, j int) int {
+		return strings.Compare(md.keys[idxes[i]], md.keys[idxes[j]])
 	})
 	return idxes
 }