Use small_vector for API iterate_args (#597)

* Use small_vector for API iterate_args

- replace dim3 value arguments with rocprofiler_dim3_t
  - dim3 has a non-trivial destructor
- common::mpl::unqualified_type
- common::stringified_argument_array_t<N> alias
- assert_public_data_type_properties()
- common::container::small_vector<T>::at function
- stringize returns small_vector<stringified_argument>
  - stack allocated vector
- remove has_pc_sampling condition (HSA, HIP)
  - this will be handled in queue interception

* Misc tweaks

[ROCm/rocprofiler-sdk commit: 8591ed1c96]

projects/rocprofiler-sdk/source/lib/common/stringize_arg.hpp

@@ -22,6 +22,7 @@
 
 #pragma once
 
+#include "lib/common/container/small_vector.hpp"
 #include "lib/common/mpl.hpp"
 
 #include <fmt/core.h>
@@ -40,10 +41,14 @@ struct stringified_argument
     int32_t     indirection_level = 0;
     int32_t     dereference_count = 0;
     const char* type              = nullptr;
-    std::string name              = {};
+    const char* name              = nullptr;
     std::string value             = {};
 };
 
+template <size_t N>
+using stringified_argument_array_t =
+    container::small_vector<stringified_argument, std::min<size_t>(N, 6)>;
+
 template <typename Tp, typename FuncT>
 auto
 stringize_arg_impl(const Tp& _v, const int32_t max_deref, int32_t& deref_cnt, FuncT&& impl)
@@ -114,7 +119,7 @@ stringize_arg(int32_t max_deref, const std::pair<const char*, Tp>& arg, FuncT&&
     auto _arg              = common::stringified_argument{};
     _arg.indirection_level = mpl::indirection_level<Tp>::value;
     _arg.type              = typeid(Tp).name();
-    _arg.name              = std::string{arg.first};
+    _arg.name              = arg.first;
     _arg.value             = stringize_arg_impl(
         arg.second, max_deref, _arg.dereference_count, std::forward<FuncT>(impl));
     return _arg;