Compile certain source files in libtensor with no-fast-math

[oleksandr-pavlyk]

Closes gh-1314 by compiling Python binding code converting float('nan') to a floating-point number using -fno-fast-math.

• Have you provided a meaningful PR description?
• Have you added a test, reproducer or referred to an issue with a reproducer?
• Have you tested your changes locally for CPU and GPU devices?
• Have you made sure that new changes do not introduce compiler warnings?
• Have you checked performance impact of proposed changes?
• If this PR is a work in progress, are you opening the PR as a draft?

---

To see that this is the right solution, use

// filename: nan.cpp
#include "pybind11/pybind11.h"

double fp_id(double x) {
  return x;
}

PYBIND11_MODULE(_nan, m) {
  m.def("fp_id", &fp_id);
}

Compile it

icpx nan.cpp --shared -fPIC $(python -m pybind11 --includes) -o _nan.so

and test with

python -c "import _nan; print(_nan.fp_id(float('nan')))"

With default compilation option -1 comes out, signaling an uncaught exception. It takes -fno-fast-math for it to round-trip correctly.

[github-actions]

View rendered docs @ https://intelpython.github.io/dpctl/pulls/1318/index.html

[coveralls]

coverage: 84.795%. remained the same when pulling 3baa622 on fix-gh-1314 into dcca136 on master.

[github-actions]

Array API standard conformance tests for dpctl=0.14.6dev0=py310h7bf5fec_70 ran successfully.
Passed: 595
Failed: 405
Skipped: 119

[github-actions]

Deleted rendered PR docs from intelpython.github.com/dpctl, latest should be updated shortly. 🤞

[github-actions]

Array API standard conformance tests for dpctl=0.14.6dev0=py310h7bf5fec_90 ran successfully.
Passed: 639
Failed: 361
Skipped: 119