Avoid closure in tree_mapreduce

[MilesCranmer]

This refactors tree_mapreduce to avoid the use of boxed variables in a closure. I don't think this will affect performance, but it's just more readable.

This also avoids the use of @with_memoization in tree_mapreduce with was quite hard to understand.

This also sets the stage for n-arity operators.

Finally, this deprecates Julia versions earlier than 1.10 which are no longer maintained by Julia.

[github-actions]

Benchmark Results

	master	1cf33ed...	master/1cf33ed0252fb5...
eval/ComplexF32/evaluation	7.3 ± 0.49 ms	7.4 ± 0.46 ms	0.987
eval/ComplexF64/evaluation	9.47 ± 0.72 ms	9.57 ± 0.74 ms	0.99
eval/Float32/derivative	10.7 ± 1.6 ms	10.9 ± 2 ms	0.988
eval/Float32/derivative_turbo	10.8 ± 1.7 ms	10.9 ± 2 ms	0.987
eval/Float32/evaluation	2.63 ± 0.24 ms	2.62 ± 0.22 ms	1
eval/Float32/evaluation_bumper	0.533 ± 0.012 ms	0.561 ± 0.014 ms	0.951
eval/Float32/evaluation_turbo	0.548 ± 0.024 ms	0.571 ± 0.029 ms	0.961
eval/Float32/evaluation_turbo_bumper	0.532 ± 0.013 ms	0.557 ± 0.013 ms	0.957
eval/Float64/derivative	14.2 ± 0.56 ms	14.7 ± 0.66 ms	0.963
eval/Float64/derivative_turbo	14.1 ± 0.56 ms	14.7 ± 0.78 ms	0.961
eval/Float64/evaluation	2.75 ± 0.24 ms	2.81 ± 0.24 ms	0.978
eval/Float64/evaluation_bumper	1.19 ± 0.043 ms	1.21 ± 0.044 ms	0.986
eval/Float64/evaluation_turbo	1.03 ± 0.056 ms	1.05 ± 0.056 ms	0.98
eval/Float64/evaluation_turbo_bumper	1.19 ± 0.042 ms	1.22 ± 0.044 ms	0.98
utils/combine_operators/break_sharing	0.0393 ± 0.0008 ms	0.0389 ± 0.00053 ms	1.01
utils/convert/break_sharing	22.6 ± 0.93 μs	23.2 ± 0.83 μs	0.978
utils/convert/preserve_sharing	0.122 ± 0.0024 ms	0.115 ± 0.0025 ms	1.06
utils/copy/break_sharing	23.3 ± 0.94 μs	24.4 ± 0.91 μs	0.956
utils/copy/preserve_sharing	0.124 ± 0.0025 ms	0.118 ± 0.0025 ms	1.05
utils/count_constant_nodes/break_sharing	11.7 ± 0.29 μs	9.02 ± 0.11 μs	1.3
utils/count_constant_nodes/preserve_sharing	0.109 ± 0.0026 ms	0.102 ± 0.0021 ms	1.07
utils/count_depth/break_sharing	12.6 ± 0.28 μs	10 ± 0.15 μs	1.26
utils/count_nodes/break_sharing	10.6 ± 0.3 μs	8.56 ± 0.11 μs	1.24
utils/count_nodes/preserve_sharing	0.109 ± 0.0024 ms	0.103 ± 0.0024 ms	1.06
utils/get_set_constants!/break_sharing	0.0342 ± 0.0012 ms	0.0344 ± 0.001 ms	0.994
utils/get_set_constants!/preserve_sharing	0.222 ± 0.0051 ms	0.211 ± 0.0054 ms	1.05
utils/get_set_constants_parametric	0.0478 ± 0.0018 ms	0.0486 ± 0.0015 ms	0.982
utils/has_constants/break_sharing	4.36 ± 0.06 μs	4.28 ± 0.055 μs	1.02
utils/has_operators/break_sharing	1.91 ± 0.023 μs	1.93 ± 0.019 μs	0.988
utils/hash/break_sharing	24.9 ± 0.63 μs	26 ± 0.56 μs	0.959
utils/hash/preserve_sharing	0.131 ± 0.0024 ms	0.122 ± 0.0029 ms	1.07
utils/index_constant_nodes/break_sharing	22.6 ± 0.73 μs	23.1 ± 0.73 μs	0.979
utils/index_constant_nodes/preserve_sharing	0.123 ± 0.0023 ms	0.115 ± 0.0023 ms	1.07
utils/is_constant/break_sharing	4.18 ± 0.066 μs	4.07 ± 0.058 μs	1.03
utils/simplify_tree/break_sharing	0.166 ± 0.0013 ms	0.174 ± 0.0014 ms	0.953
utils/simplify_tree/preserve_sharing	0.279 ± 0.0046 ms	0.276 ± 0.004 ms	1.01
utils/string_tree/break_sharing	0.536 ± 0.013 ms	0.536 ± 0.011 ms	1
utils/string_tree/preserve_sharing	0.664 ± 0.013 ms	0.661 ± 0.022 ms	1
time_to_load	0.242 ± 0.0043 s	0.263 ± 0.0015 s	0.923

[coveralls]

Pull Request Test Coverage Report for Build 11300259073

Details

• 20 of 20 (100.0%) changed or added relevant lines in 1 file are covered.
• 20 unchanged lines in 4 files lost coverage.
• Overall coverage decreased (-0.5%) to 95.357%

---

Files with Coverage Reduction	New Missed Lines	%
src/OperatorEnumConstruction.jl	1	95.19%
src/precompile.jl	2	87.5%
src/EvaluateDerivative.jl	4	88.0%
src/Evaluate.jl	13	94.75%

Totals
Change from base Build 11221646120:	-0.5%
Covered Lines:	2403
Relevant Lines:	2520

---

💛 - Coveralls