Do not use link-huge-device-code option

[oleksandr-pavlyk]

This PR remove use of -fsycl-link-huge-device-code link option when building tensor_impl modules on Linux.

• 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?

[github-actions]

View rendered docs @ https://intelpython.github.io/dpctl/pulls/1497/index.html

[coveralls]

coverage: 90.187%. remained the same
when pulling a4c17c9 on get-rid-of-huge-device-code
into fbe0081 on master.

[github-actions]

Array API standard conformance tests for dpctl=0.15.1dev3=py310h15de555_7 ran successfully.
Passed: 891
Failed: 18
Skipped: 86

[ndgrigorian]

According to the llvm docs the option was deprecated in favor of -flink-huge-device-code, and according to release notes this option is mostly identical. Would it make more sense to use the replacement?

[oleksandr-pavlyk]

It is being removed not because of the deprecation warning, but because it is no longer necessary after splitting single monolithic _tensor_impl into a collection of smaller modules, and suspectedly makes linking of device code somewhat longer.

[ndgrigorian]

I see, then in that case, this LGTM.