Move SymbolicUtils.jl into extension

[MilesCranmer]

This makes it so you need to call:

using SymbolicUtils
using DynamicExpressions

for the functions node_to_symbolic and symbolic_to_node to be loaded. I think this makes sense because you wouldn't typically use these without also having this package set up.

This allows downstream applications to reduce the number of dependencies if the algebraic simplification is not needed.

[github-actions[bot]]

Pull Request Test Coverage Report for Build 5093184845

• 0 of 4 (0.0%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+2.4%) to 90.811%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
src/ExtensionInterface.jl	0	4	0.0%
<!--	Total:	0	4	0.0%	-->

Totals
Change from base Build 4958507899:	2.4%
Covered Lines:	840
Relevant Lines:	925

---

💛 - Coveralls

[github-actions[bot]]

Benchmark Results

	master	42311a60050a94...	t[master]/t[42311a60050a94...]
eval/ComplexF32/evaluation	12.3 ± 1.1 ms	12.5 ± 1.2 ms	0.979
eval/ComplexF64/evaluation	15.6 ± 1.9 ms	15.6 ± 1.6 ms	1
eval/Float32/derivative	23.6 ± 3.4 ms	23 ± 3 ms	1.03
eval/Float32/derivative_turbo	23.5 ± 3.3 ms	23.4 ± 3.3 ms	1.01
eval/Float32/evaluation	4.35 ± 0.53 ms	4.34 ± 0.55 ms	1
eval/Float32/evaluation_turbo	1.33 ± 0.23 ms	1.33 ± 0.24 ms	1
eval/Float64/derivative	29.9 ± 2.5 ms	31.3 ± 2.5 ms	0.955
eval/Float64/derivative_turbo	0.0339 ± 0.0025 s	0.0347 ± 0.0032 s	0.977
eval/Float64/evaluation	4.58 ± 0.6 ms	4.64 ± 0.63 ms	0.986
eval/Float64/evaluation_turbo	2.52 ± 0.43 ms	2.52 ± 0.38 ms	0.998
time_to_load	1.97 ± 0.15 s	2.05 ± 0.04 s	0.964
utils/combine_operators/break_sharing	0.0578 ± 0.014 ms	0.0656 ± 0.023 ms	0.881
utils/convert/break_sharing	0.0591 ± 0.018 ms	0.0548 ± 0.017 ms	1.08
utils/convert/preserve_sharing	0.24 ± 0.062 ms	0.239 ± 0.072 ms	1.01
utils/copy/break_sharing	0.0518 ± 0.015 ms	0.0498 ± 0.014 ms	1.04
utils/copy/preserve_sharing	0.21 ± 0.05 ms	0.204 ± 0.047 ms	1.03
utils/count_constants/break_sharing	26.5 ± 5.5 μs	24.5 ± 4.8 μs	1.08
utils/count_depth/break_sharing	28.6 ± 6 μs	29.5 ± 7.1 μs	0.97
utils/count_nodes/break_sharing	25 ± 4.6 μs	24.7 ± 6.1 μs	1.01
utils/get_set_constants!/break_sharing	0.0892 ± 0.02 ms	0.0916 ± 0.023 ms	0.974
utils/has_constants/break_sharing	7.67 ± 1.5 μs	7.7 ± 1.5 μs	0.996
utils/has_operators/break_sharing	1.98 ± 0.23 μs	2.48 ± 0.36 μs	0.798
utils/index_constants/break_sharing	0.0752 ± 0.016 ms	0.0852 ± 0.027 ms	0.883
utils/is_constant/break_sharing	7.7 ± 1.3 μs	7.86 ± 2.3 μs	0.98
utils/simplify_tree/break_sharing	0.151 ± 0.033 ms	0.155 ± 0.034 ms	0.98

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).