Update Intel compiler from icc to icx

[hpkfft]

The Intel(R) oneAPI DPC++/C++ Compiler 2026.0.0 is not officially supported by CPython, and I am not suggesting to change that.
However, I see that there is code in configure.ac that applies to the Intel classic compiler (icc for C and icpc for C++). The last version of the classic compiler was 2021.10 and it stopped shipping in 2023.

I would like to suggest updating icc to icx in configure.ac.

Note that not all tests pass. In particular:

  File "/tmp/Lib/test/test_cmath.py", line 412, in test_phase
    self.assertAlmostEqual(phase(0), 0.)
                           ~~~~~^^^
ValueError: math domain error

and

  File "/tmp/Lib/test/test_math.py", line 304, in testAtan2
    self.ftest('atan2(0., -0.)', math.atan2(0., -0.), math.pi)
                                 ~~~~~~~~~~^^^^^^^^^
ValueError: math domain error

I do not think the Intel math library is wrong here; it's OK to set errno to EDOM for atan2(0.0, 0.0). GCC using glibc does not, and that's OK too. Both libraries return 0.0 for atan2(0.0, 0.0) and math.pi for atan2(0.0, -0.0).

If I comment out those tests, I can compile locally with PGO. Also, icx does support -Wno-unused-result, and -ffp-model=precise is needed or else other floating-point tests will fail. (As noted in the comment, clang uses precise by default. The Intel compiler is based on clang, but it has changed the default to fast.)

[bedevere-app]

Most changes to Python require a NEWS entry. Add one using the blurb_it web app or the blurb command-line tool.

If this change has little impact on Python users, wait for a maintainer to apply the skip news label instead.

[bedevere-app]

Most changes to Python require a NEWS entry. Add one using the blurb_it web app or the blurb command-line tool.

If this change has little impact on Python users, wait for a maintainer to apply the skip news label instead.

[StanFromIreland]

Please open an issue.