[libc] Disable the string_to_float test on NVPTX

[jhuber6]

Summary

This test began failing after recent changes. Disable it for now.

[michaelrj]

The patch is fine, but I am confused as to how the str_to_float tests were passing with the high_precision_decimal tests failing, given that str_to_float uses high_precision_decimal.

[jhuber6]

In D152032#4391807, @michaelrj wrote:

The patch is fine, but I am confused as to how the str_to_float tests were passing with the high_precision_decimal tests failing, given that str_to_float uses high_precision_decimal.

I'm not sure either. Haven't had the time to properly diagnose these failings on NVPTX. It's also more difficult for me since it's not my primary development system. I have a feeling there might be a common cause between the spurious failures on NVPTX here but I still to sit down and try to make a reproducer.

libc/test/src/__support/CMakeLists.txt

add_libc_test(
   SUITE
     libc-support-tests
   SRCS
     endian_test.cpp
   DEPENDS
     libc.src.__support.common
 )
 
-# This test fails with a segmentation fault on NVPTX.
+# These tests fails with an illegal instruction
 if(NOT LIBC_GPU_TARGET_ARCHITECTURE_IS_NVPTX)
   add_libc_test(
     high_precision_decimal_test
     SUITE
       libc-support-tests
     SRCS
     high_precision_decimal_test.cpp
     DEPENDS
       libc.src.__support.high_precision_decimal
       libc.src.__support.uint128
   )
-endif()
 
   add_libc_test(
     str_to_float_test
     SUITE
       libc-support-tests
     SRCS
       str_to_float_test.cpp
     DEPENDS
       libc.src.__support.str_to_float
       libc.src.__support.uint128
       libc.src.errno.errno
   )
+endif()
 
 add_libc_test(
   integer_to_string_test
   SUITE
     libc-support-tests
   SRCS
     integer_to_string_test.cpp
   DEPENDS