benchtests: Add benchmarks for frexp functions

Add benchmark support for frexp, frexpf, and frexpl to measure the
performance improvement of the fast path optimization.

- Created frexp-inputs, frexpf-inputs, frexpl-inputs with random test values
- Added frexp, frexpf, frexpl to bench-math list
- Added CFLAGS to disable builtins for accurate benchmarking

These benchmarks will be used to quantify the performance gains from the
fast path optimization for normal floating-point numbers.

Signed-off-by: Osama Abdelkader <osama.abdelkader@gmail.com>

benchtests/Makefile

@@ -82,6 +82,9 @@ bench-math := \
   fminf \
   fmod \
   fmodf \
+  frexp \
+  frexpf \
+  frexpl \
   hypot \
   hypotf \
   ilogb \
@@ -347,6 +350,9 @@ CFLAGS-bench-trunc.c += -fno-builtin
 CFLAGS-bench-truncf.c += -fno-builtin
 CFLAGS-bench-fmod.c += -fno-builtin
 CFLAGS-bench-fmodf.c += -fno-builtin
+CFLAGS-bench-frexp.c += -fno-builtin
+CFLAGS-bench-frexpf.c += -fno-builtin
+CFLAGS-bench-frexpl.c += -fno-builtin
 CFLAGS-bench-remainder.c += -fno-builtin
 CFLAGS-bench-remainderf.c += -fno-builtin
 CFLAGS-bench-roundeven.c += -fno-builtin