for >3 loops in a group, don't permute all possible orders

[chriselrod]

This didn't appear to help compile time performance in the case I was interested in.

[codecov]

Codecov Report

Merging #398 (dc53c94) into main (7a0ffd4) will increase coverage by 0.39%.
The diff coverage is 100.00%.

❗ Current head dc53c94 differs from pull request most recent head 608c66d. Consider uploading reports for the commit 608c66d to get more accurate results

@@            Coverage Diff             @@
##             main     #398      +/-   ##
==========================================
+ Coverage   86.95%   87.34%   +0.39%     
==========================================
  Files          37       38       +1     
  Lines        9305     9545     +240     
==========================================
+ Hits         8091     8337     +246     
+ Misses       1214     1208       -6     

Impacted Files	Coverage Δ
src/modeling/determinestrategy.jl	95.62% <100.00%> (+0.67%)	⬆️
src/user_api_conveniences.jl	90.47% <0.00%> (-9.53%)	⬇️
src/codegen/lower_threads.jl	48.93% <0.00%> (-4.23%)	⬇️
src/simdfunctionals/map.jl	88.50% <0.00%> (-4.07%)	⬇️
src/simdfunctionals/vmap_grad_rrule.jl	90.62% <0.00%> (-1.57%)	⬇️
src/codegen/lower_memory_common.jl	86.64% <0.00%> (-0.37%)	⬇️
src/condense_loopset.jl	96.89% <0.00%> (-0.20%)	⬇️
src/codegen/lowering.jl	92.60% <0.00%> (-0.08%)	⬇️
src/constructors.jl	98.72% <0.00%> (ø)
src/getconstindexes.jl	100.00% <0.00%> (ø)
... and 11 more

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