[WIP] Commutative operations in rules

[Bumblebee00]

this pr builds on top of defslot and negative exponent matching. Before this pr julia symbolic rewriting rules had this unfortunate behaviour:

julia> @syms a x
(a, x)

julia> rule = @rule sin(~x) + cos(~x) => "yes"
sin(~x) + cos(~x) => 1

julia> arguments(sin(x) + cos(x))
2-element SymbolicUtils.SmallVec{Any, Vector{Any}}:
 cos(x)
 sin(x)

julia> rule(sin(x)+cos(x)) # doesnt match bc arguments are ordered cos first and then sin


julia> arguments(sin(a) + cos(a))
2-element SymbolicUtils.SmallVec{Any, Vector{Any}}:
 sin(a)
 cos(a)

julia> rule(sin(a)+cos(a)) # match bc arguments are ordered sin first and then cos
"yes"

there was the acrule macro that helped, but was working only on the first operation of the expression, not every level. with this pr instead every * and + operation creates a matcher that checks all possible orderings of its arguments.

This can also be a bad thing becuse the cases to check could be too many...

Also I refacored the code from my last two pr to make everything clearer and cleaner

[github-actions]

Benchmark Results

	master	70e36ed...	master / 70e36ed...
overhead/acrule/a+2	0.937 ± 0.27 μs	1.02 ± 0.024 μs	0.917 ± 0.26
overhead/acrule/a+2+b	0.925 ± 0.23 μs	0.979 ± 0.015 μs	0.945 ± 0.23
overhead/acrule/a+b	0.262 ± 0.025 μs	0.26 ± 0.095 μs	1.01 ± 0.38
overhead/acrule/noop:Int	1.56 ± 0.01 ns	1.56 ± 0.01 ns	1 ± 0.0091
overhead/acrule/noop:Sym	28.2 ± 6.4 ns	16.2 ± 2.8 ns	1.74 ± 0.5
overhead/get_degrees/large_poly	0.156 ± 0.0087 s	0.445 ± 0.038 s	0.351 ± 0.036
overhead/rule/noop:Int	0.0363 ± 0.024 μs	0.0349 ± 0.023 μs	1.04 ± 0.97
overhead/rule/noop:Sym	0.035 ± 0.023 μs	0.034 ± 0.024 μs	1.03 ± 0.99
overhead/rule/noop:Term	0.0353 ± 0.023 μs	0.0341 ± 0.024 μs	1.04 ± 1
overhead/ruleset/noop:Int	0.118 ± 0.036 μs	0.115 ± 0.035 μs	1.02 ± 0.44
overhead/ruleset/noop:Sym	0.127 ± 0.04 μs	0.126 ± 0.041 μs	1.01 ± 0.45
overhead/ruleset/noop:Term	5.2 ± 0.59 μs	4.6 ± 0.47 μs	1.13 ± 0.17
overhead/simplify/noop:Int	0.32 ± 0.0088 μs	0.246 ± 0.035 μs	1.3 ± 0.19
overhead/simplify/noop:Sym	0.324 ± 0.0086 μs	0.241 ± 0.034 μs	1.35 ± 0.19
overhead/simplify/noop:Term	0.0506 ± 0.0056 ms	0.089 ± 0.0099 ms	0.569 ± 0.089
overhead/simplify/randterm (+, *):serial	0.124 ± 0.011 s	0.388 ± 0.026 s	0.32 ± 0.035
overhead/simplify/randterm (+, *):thread	0.0706 ± 0.0059 s	0.255 ± 0.035 s	0.276 ± 0.045
overhead/simplify/randterm (/, *):serial	0.248 ± 0.023 ms	0.286 ± 0.025 ms	0.867 ± 0.11
overhead/simplify/randterm (/, *):thread	0.281 ± 0.027 ms	0.319 ± 0.029 ms	0.879 ± 0.12
overhead/substitute/a	0.106 ± 0.011 ms	0.104 ± 0.012 ms	1.02 ± 0.16
overhead/substitute/a,b	0.0896 ± 0.0094 ms	0.0906 ± 0.0074 ms	0.989 ± 0.13
overhead/substitute/a,b,c	20.2 ± 1.8 μs	20.1 ± 2.1 μs	1 ± 0.14
polyform/easy_iszero	0.0422 ± 0.006 ms	0.0403 ± 0.0011 ms	1.05 ± 0.15
polyform/isone	3.71 ± 0.92 ns	3.1 ± 0.01 ns	1.19 ± 0.3
polyform/iszero	1.41 ± 0.037 ms	1.41 ± 0.038 ms	1 ± 0.038
polyform/simplify_fractions	1.92 ± 0.064 ms	1.93 ± 0.083 ms	0.995 ± 0.054
time_to_load	1.13 ± 0.0087 s	1.14 ± 0.0086 s	0.992 ± 0.011

Benchmark Plots

A plot of the benchmark results have been uploaded as an artifact to the workflow run for this PR.
Go to "Actions"->"Benchmark a pull request"->[the most recent run]->"Artifacts" (at the bottom).