Replace __init__ with OncePerProcess on supported Julia versions

[fredrikekre]

No description provided.

[codecov]

Codecov Report

Attention: Patch coverage is 84.21053% with 6 lines in your changes missing coverage. Please review.

Project coverage is 71.50%. Comparing base (cefd827) to head (0749489).
Report is 1 commits behind head on master.

Files with missing lines	Patch %	Lines
src/fft.jl	62.50%	6 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #316      +/-   ##
==========================================
+ Coverage   70.99%   71.50%   +0.50%     
==========================================
  Files           5        5              
  Lines         531      544      +13     
==========================================
+ Hits          377      389      +12     
- Misses        154      155       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[fredrikekre]

CC @giordano, would you be able to review and release this change if it looks ok? Thanks!

[giordano]

For what is worth, with FFTW.jl v1.8.1:

% for i in {1..5}; do julia +nightly --project=/tmp -e '@time using FFTW'; done
  0.057132 seconds (69.55 k allocations: 4.520 MiB, 4.67% compilation time)
  0.056803 seconds (69.55 k allocations: 4.520 MiB, 4.87% compilation time)
  0.056254 seconds (69.55 k allocations: 4.520 MiB, 4.78% compilation time)
  0.059455 seconds (69.55 k allocations: 4.520 MiB, 4.53% compilation time)
  0.056675 seconds (69.55 k allocations: 4.520 MiB, 4.71% compilation time)

On dbcb5d2

% for i in {1..5}; do julia +nightly --project=/tmp -e '@time using FFTW'; done
  0.053584 seconds (69.61 k allocations: 4.526 MiB)
  0.053191 seconds (69.61 k allocations: 4.526 MiB)
  0.053056 seconds (69.61 k allocations: 4.526 MiB)
  0.053553 seconds (69.61 k allocations: 4.526 MiB)
  0.053436 seconds (69.61 k allocations: 4.526 MiB)

Not much faster, but it seems to be consistently faster.