Perform in-place FFT in intermediate steps of ND FFT

[vtavana]

In this PR, FFT module is updated to perform in-place FFT in intermediate steps of ND FFT which brings performance improvements for cases such as:

import dpnp
shape = (30,40,50,60,70)
a = dpnp.random.rand(*shape) + 1j * dpnp.random.rand(*shape)
res = dpnp.fft.fftn(a)
%timeit res = dpnp.fft.fftn(a); res.sycl_queue.wait()
# 74.6 ms ± 166 μs per loop (mean ± std. dev. of 7 runs, 10 loops each)# Master branch
# 71.7 ms ± 99.8 μs per loop (mean ± std. dev. of 7 runs, 10 loops each) # This branch

import dpnp
shape = (30,40,50,60,70)
a = dpnp.random.rand(*shape)
out = out=dpnp.empty((30, 40, 50, 60, 36), dtype=dpnp.complex128)
res = dpnp.fft.rfftn(a, out=out)
res is out
# True
%timeit res = dpnp.fft.rfftn(a, out=out); res.sycl_queue.wait()
# 78.8 ms ± 384 μs per loop (mean ± std. dev. of 7 runs, 10 loops each) # Master branch
# 69.6 ms ± 86.1 μs per loop (mean ± std. dev. of 7 runs, 10 loops each) # This branch

• 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?
• Have you added documentation for your changes, if necessary?
• Have you added your changes to the changelog?

[github-actions]

View rendered docs @ https://intelpython.github.io/dpnp/pull/2543/index.html

[github-actions]

Array API standard conformance tests for dpnp=0.19.0dev1=py313h509198e_27 ran successfully.
Passed: 1227
Failed: 0
Skipped: 9

[coveralls]

coverage: 72.108% (+0.007%) from 72.101%
when pulling b9c79c5 on fft-out
into c8b9ad0 on master.

[antonwolfy]

Thank you @vtavana for implementing a significant performance improvement. LGTM!