Use SizedUnsignedType_t to pick UnsignedType for uniform_integer_distribution

The previously used `make_unsigned` approach combined with the overload
set of `extendedMult` caused compilation issues on MacOS platform. By
forcing the selection to be one of `std::uintX_t` types we don't need
to complicate the overload set further.

src/catch2/internal/catch_uniform_integer_distribution.hpp

@@ -13,22 +13,6 @@
 
 namespace Catch {
 
-    namespace Detail {
-        // Indirection to enable make_unsigned<bool> behaviour.
-        template <typename T>
-        struct make_unsigned {
-            using type = std::make_unsigned_t<T>;
-        };
-
-        template <>
-        struct make_unsigned<bool> {
-            using type = uint8_t;
-        };
-
-        template <typename T>
-        using make_unsigned_t = typename make_unsigned<T>::type;
-    }
-
 /**
  * Implementation of uniform distribution on integers.
  *
@@ -44,7 +28,7 @@ template <typename IntegerType>
 class uniform_integer_distribution {
     static_assert(std::is_integral<IntegerType>::value, "...");
 
-    using UnsignedIntegerType = Detail::make_unsigned_t<IntegerType>;
+    using UnsignedIntegerType = Detail::SizedUnsignedType_t<sizeof(IntegerType)>;
 
     // Only the left bound is stored, and we store it converted to its
     // unsigned image. This avoids having to do the conversions inside

tests/SelfTest/IntrospectiveTests/RandomNumberGeneration.tests.cpp

@@ -140,7 +140,9 @@ TEMPLATE_TEST_CASE( "uniform_integer_distribution can handle unit ranges",
                     uint32_t,
                     int32_t,
                     uint64_t,
-                    int64_t ) {
+                    int64_t,
+                    size_t,
+                    ptrdiff_t) {
     // We want random seed to sample different parts of the rng state,
     // the output is predetermined anyway
     std::random_device rd;