Explicitly include headers used in the c++ file pertaining to reduction

[oleksandr-pavlyk]

This is a good, recommended practice.

Moved global definition of namespaces td_ns and su_ns in 'reductions.hpp' into namespace.

Fixed resulting build breakage by including needed headers and defining those namespaces locally in affected files.

Changed definition of logaddexp in math_utils.hpp to use selector instead of if/else branches to improve performance. Switch revealed silent inefficiency from using std::log1p for all types (for sycl::half it would silently upcast to float and evaluate std::log1p<float>(sycl::half)).

Used if constexpr to use sycl::log1p for half and single FP types. For double, use std::log1p until the issue is fixed.

• 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]

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

[github-actions]

Array API standard conformance tests for dpctl=0.18.0dev0=py310hdf72452_324 ran successfully.
Passed: 894
Failed: 1
Skipped: 119

[coveralls]

coverage: 87.91%. remained the same
when pulling 146f66e on clean-up-reduction-headers
into 0b9cec8 on master.

[ndgrigorian]

Test failures are unrelated, this LGTM.

Good catch on the namespaces!