[flang-rt] Simplify INDEX with len-1 SUBSTRING. (#137889)

The len-1 case is noticeably slower than gfortran's straightforward
implementation
https://github.com/gcc-mirror/gcc/blob/075611b646e5554ae02b2622061ea1614bf16ead/libgfortran/intrinsics/string_intrinsics_inc.c#L253
This change speeds up a simple microkernel by 37% on icelake.

Author: vzakhari

flang-rt/lib/runtime/character.cpp

@@ -14,6 +14,7 @@
 #include "flang/Common/uint128.h"
 #include "flang/Runtime/character.h"
 #include "flang/Runtime/cpp-type.h"
+#include "flang/Runtime/freestanding-tools.h"
 #include <algorithm>
 #include <cstring>
 
@@ -293,6 +294,24 @@ inline RT_API_ATTRS std::size_t Index(const CHAR *x, std::size_t xLen,
     }
     return 0;
   }
+  if (wantLen == 1) {
+    // Trivial case for single character lookup.
+    // We can use simple forward search.
+    CHAR ch{want[0]};
+    if constexpr (std::is_same_v<CHAR, char>) {
+      if (auto pos{reinterpret_cast<const CHAR *>(
+              Fortran::runtime::memchr(x, ch, xLen))}) {
+        return pos - x + 1;
+      }
+    } else {
+      for (std::size_t at{0}; at < xLen; ++at) {
+        if (x[at] == ch) {
+          return at + 1;
+        }
+      }
+    }
+    return 0;
+  }
   // Non-trivial forward substring search: use a simplified form of
   // Boyer-Moore substring searching.
   for (std::size_t at{1}; at + wantLen - 1 <= xLen;) {

flang-rt/unittests/Runtime/CharacterTest.cpp

@@ -354,6 +354,7 @@ TYPED_TEST(SearchTests, IndexTests) {
       {"", "a", true, 0},
       {"aa", "a", false, 1},
       {"aa", "a", true, 2},
+      {"aAA", "A", false, 2},
       {"Fortran that I ran", "that I ran", false, 9},
       {"Fortran that I ran", "that I ran", true, 9},
       {"Fortran that you ran", "that I ran", false, 0},