Optimize the iteration scheme in CliffordNumbers.mul for performance

[brainandforce]

There is a significant discrepancy in performance on hardware with SIMD if two differently-sized arguments are multliplied in reverse order: in general, larger first arguments result in slower runtimes because there are more loop iterations as opposed to vectorized operations. This change generates the syntax tree differently depending on argument size.

Perhaps in the future we can come up with a heuristic to optimize code generation even further. Additionally, I want to unify the cases so we don't have a lot of boilerplate code.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 85.37%. Comparing base (fce1003) to head (b6af29b).

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #12      +/-   ##
==========================================
+ Coverage   84.19%   85.37%   +1.17%     
==========================================
  Files          14       14              
  Lines         696      711      +15     
==========================================
+ Hits          586      607      +21     
+ Misses        110      104       -6     

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