gh-112535: Update _Py_ThreadId() to support RISC-V

[furkanonder]

• Issue: Make _Py_ThreadId() work on PowerPC, IBM Z, etc. #112535

• Clang supports __builtin_thread_pointer for RISC-V from clang 11.0.0
• GCC supports __builtin_thread_pointer for RISC-V from GCC 10.2.0 (The Clang team said that checking the __builtin_thread_pointer with __has_builtin but not safe for GCC)

[furkanonder]

GCC

 ./configure --disable-gil --with-pydebug

Test:

user@starfive:~/cpython$ ./python -m test -j3 

Output:

== Tests result: FAILURE ==
12 tests skipped:
test.test_asyncio.test_windows_events
test.test_asyncio.test_windows_utils test_devpoll test_ioctl
test_kqueue test_launcher test_msvcrt test_perf_profiler
test_startfile test_winconsoleio test_winreg test_wmi
10 tests skipped (resource denied):
test_curses test_peg_generator test_smtpnet test_socketserver
test_tkinter test_ttk test_urllib2net test_urllibnet test_winsound
test_zipfile64
1 test failed:
test_cmd_line
446 tests OK.
Total duration: 35 min 12 sec
Total tests: run=40,908 failures=2 skipped=1,401
Total test files: run=459/469 failed=1 skipped=12 resource_denied=10
Result: FAILURE

The failed test_cmd_line test was already an existing issue for RISC-V.

[furkanonder]

CLANG

./configure --disable-gil --with-pydebug --with-cxx-main=/usr/bin/clang++ CC=/usr/bin/clang

Test:

user@starfive:~/cpython$ ./python -m test -j3 

Output:

== Tests result: FAILURE ==
12 tests skipped:
test.test_asyncio.test_windows_events
test.test_asyncio.test_windows_utils test_devpoll test_ioctl
test_kqueue test_launcher test_msvcrt test_perf_profiler
test_startfile test_winconsoleio test_winreg test_wmi
10 tests skipped (resource denied):
test_curses test_peg_generator test_smtpnet test_socketserver
test_tkinter test_ttk test_urllib2net test_urllibnet test_winsound
test_zipfile64
3 tests failed:
test.test_gdb.test_backtrace test.test_gdb.test_pretty_print
test_cmd_line
444 tests OK.
Total duration: 31 min 45 sec
Total tests: run=40,908 failures=24 skipped=1,415
Total test files: run=459/469 failed=3 skipped=12 resource_denied=10
Result: FAILURE

The failed test_cmd_line test and test.test_gdb.test_backtrace, test.test_gdb.test_pretty_print were already existing issues for RISC-V.

[corona10]

https://gcc.godbolt.org/z/TYe716Wa7

I cross-checked the assembly code from Godbolt, too.
(Both 32bit and 64bit)

[corona10]

LGTM but waiting @colesbury too.