Prevent brace initialization of AlphaNum

This was not intended to be supported, and it has resulted in calls as `absl::StrCat({...})`, which are not supported and only work coincidentally for the first 4 arguments due to `absl::StrCat` having overloads that take `absl::AlphaNum` directly for those.

The existing situation prevents modifying the implementations of such functions to alternatives that do not have such overloads for those arguments.

PiperOrigin-RevId: 599872755
Change-Id: I02c90119b2b96a922cf7e3b5d5f02affe24a272d

absl/strings/str_cat.h

@@ -93,6 +93,7 @@
 #include <cstddef>
 #include <cstddef>
 #include <cstdint>
 #include <cstdint>
 #include <cstring>
 #include <cstring>
+#include <initializer_list>
 #include <string>
 #include <string>
 #include <type_traits>
 #include <type_traits>
 #include <utility>
 #include <utility>
@@ -312,6 +313,10 @@ class AlphaNum {
   // No bool ctor -- bools convert to an integral type.
   // No bool ctor -- bools convert to an integral type.
   // A bool ctor would also convert incoming pointers (bletch).
   // A bool ctor would also convert incoming pointers (bletch).
 
 
+  // Prevent brace initialization
+  template <typename T>
+  AlphaNum(std::initializer_list<T>) = delete;  // NOLINT(runtime/explicit)
+
   AlphaNum(int x)  // NOLINT(runtime/explicit)
   AlphaNum(int x)  // NOLINT(runtime/explicit)
       : piece_(digits_, static_cast<size_t>(
       : piece_(digits_, static_cast<size_t>(
                             numbers_internal::FastIntToBuffer(x, digits_) -
                             numbers_internal::FastIntToBuffer(x, digits_) -