NaNMath.pow fix

[contradict]

NaNMath.pow needs a dispatch to prevent stack overflow with symbolic arguments.

[github-actions]

Benchmark Results

	master	fac570d...	t[master]/t[fac570d...]
overhead/acrule/a+2	1.1 ± 0.27 μs	1.08 ± 0.27 μs	1.02
overhead/acrule/a+2+b	1.1 ± 0.27 μs	1.07 ± 0.27 μs	1.03
overhead/acrule/a+b	0.372 ± 0.0086 μs	0.36 ± 0.0095 μs	1.03
overhead/acrule/noop:Int	26.2 ± 0.8 ns	26.3 ± 0.6 ns	0.997
overhead/acrule/noop:Sym	0.0546 ± 0.009 μs	0.0471 ± 0.0078 μs	1.16
overhead/rule/noop:Int	0.0485 ± 0.002 μs	0.0454 ± 0.0016 μs	1.07
overhead/rule/noop:Sym	0.0594 ± 0.0031 μs	0.0589 ± 0.0031 μs	1.01
overhead/rule/noop:Term	0.0594 ± 0.003 μs	0.0589 ± 0.0029 μs	1.01
overhead/ruleset/noop:Int	0.167 ± 0.0029 μs	0.166 ± 0.0022 μs	1
overhead/ruleset/noop:Sym	0.198 ± 0.0068 μs	0.2 ± 0.0081 μs	0.987
overhead/ruleset/noop:Term	4.27 ± 0.37 μs	4.27 ± 0.37 μs	1
overhead/simplify/noop:Int	0.241 ± 0.0087 μs	0.227 ± 0.0068 μs	1.07
overhead/simplify/noop:Sym	0.247 ± 0.0055 μs	0.249 ± 0.0067 μs	0.995
overhead/simplify/noop:Term	0.0456 ± 0.0047 ms	0.0446 ± 0.004 ms	1.02
overhead/simplify/randterm (+, *):serial	0.169 ± 0.0016 s	0.167 ± 0.0018 s	1.01
overhead/simplify/randterm (+, *):thread	0.118 ± 0.02 s	0.117 ± 0.02 s	1.01
overhead/simplify/randterm (/, *):serial	0.288 ± 0.0098 ms	0.281 ± 0.01 ms	1.02
overhead/simplify/randterm (/, *):thread	0.329 ± 0.013 ms	0.331 ± 0.013 ms	0.996
overhead/substitute/a	0.0743 ± 0.0021 ms	0.0743 ± 0.0018 ms	1
overhead/substitute/a,b	0.0652 ± 0.002 ms	0.0646 ± 0.0019 ms	1.01
overhead/substitute/a,b,c	23.1 ± 0.8 μs	22.2 ± 0.8 μs	1.04
polyform/easy_iszero	0.0401 ± 0.0037 ms	0.0405 ± 0.004 ms	0.99
polyform/isone	3.4 ± 0 ns	3.1 ± 0.1 ns	1.1
polyform/iszero	2.46 ± 0.046 ms	2.49 ± 0.042 ms	0.988
polyform/simplify_fractions	3.32 ± 0.063 ms	3.38 ± 0.062 ms	0.985
time_to_load	5.63 ± 0.036 s	5.52 ± 0.027 s	1.02

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).

[contradict]

I created separate PR's to fix unrelated errors in Documentation (#558) and DataDrivenDiffEq (#559).

I have no idea how to fix the fuzz test error but it doesn't seem related.

I'm looking at the ModelingToolkit failure, but it doesn't look related.