Optimize __rem_pio2_large

[MiloLug]

We need to process fq twice, but each subsequent item depends on previous one. That's why you previously were using two loops.

But it can be optimized to use only one loop:

For example, given jz = 10
At first, we process items # (in pairs):

• (10, 9)
• (9, 8) Now we can run simple loop that will process those pairs independently:
• (10, 9), (8, 7)
• (9, 8), (7, 6)
• (8, 7), (6, 5)
• so on...

This way, we can process each pair twice with only one loop

[github-actions]

Code size report:

   bare-arm:    +0 +0.000% 
minimal x86:    +0 +0.000% 
   unix x64:    +0 +0.000% standard
      stm32:    +0 +0.000% PYBV10
        rp2:    +0 +0.000% PICO

[codecov-commenter]

Codecov Report

Merging #11317 (cf5709d) into master (9e6885a) will not change coverage.
The diff coverage is n/a.

📣 This organization is not using Codecov’s GitHub App Integration. We recommend you install it so Codecov can continue to function properly for your repositories. Learn more

@@           Coverage Diff           @@
##           master   #11317   +/-   ##
=======================================
  Coverage   98.50%   98.50%           
=======================================
  Files         155      155           
  Lines       20549    20549           
=======================================
  Hits        20241    20241           
  Misses        308      308           

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[dpgeorge]

Thanks for the contribution.

I'd be reluctant to modify this code, because it's taken from the musl project and not developed here.

Also, it's not clear that this is an optimisation because there is still 2 loops in the second "else". Furthermore, we usually optimise for small code size, rather than performance. As such, it would be important to show that this does actually make the code faster, and that that improvement in speed is worth the cost of increased code size.