Enforce `Vararg` specialization

[MilesCranmer]

I didn't realize this but apparently Julia doesn't actually specialize functions based on variable numbers of arguments.

There's a way to get around this as described here: https://docs.julialang.org/en/v1/manual/performance-tips/#Be-aware-of-when-Julia-avoids-specializing.

This PR implements this forced specialization which should hopefully give some speedups.

[coveralls]

Pull Request Test Coverage Report for Build 7394655453

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 94.279%

---

Totals
Change from base Build 7337169551:	0.0%
Covered Lines:	1269
Relevant Lines:	1346

---

💛 - Coveralls

[github-actions[bot]]

Benchmark Results

	master	a16288c604ab94...	t[master]/t[a16288c604ab94...]
eval/ComplexF32/evaluation	7.38 ± 0.46 ms	7.34 ± 0.46 ms	1.01
eval/ComplexF64/evaluation	9.65 ± 0.7 ms	9.58 ± 0.73 ms	1.01
eval/Float32/derivative	10.5 ± 1.4 ms	10.8 ± 1.8 ms	0.977
eval/Float32/derivative_turbo	11.9 ± 1.4 ms	12.2 ± 1.7 ms	0.981
eval/Float32/evaluation	2.66 ± 0.22 ms	2.69 ± 0.23 ms	0.991
eval/Float32/evaluation_turbo	0.653 ± 0.034 ms	0.62 ± 0.027 ms	1.05
eval/Float64/derivative	13.8 ± 0.98 ms	13.7 ± 0.57 ms	1
eval/Float64/derivative_turbo	14.4 ± 0.67 ms	14.3 ± 0.65 ms	1
eval/Float64/evaluation	2.81 ± 0.24 ms	2.83 ± 0.24 ms	0.992
eval/Float64/evaluation_turbo	1.12 ± 0.061 ms	1.11 ± 0.06 ms	1.01
utils/combine_operators/break_sharing	0.0405 ± 0.0027 ms	0.0409 ± 0.0027 ms	0.989
utils/convert/break_sharing	28.5 ± 0.61 μs	28.1 ± 0.61 μs	1.01
utils/convert/preserve_sharing	0.13 ± 0.0035 ms	0.128 ± 0.0028 ms	1.02
utils/copy/break_sharing	29.8 ± 1.5 μs	28.8 ± 0.62 μs	1.04
utils/copy/preserve_sharing	0.129 ± 0.0032 ms	0.129 ± 0.0027 ms	0.999
utils/count_constants/break_sharing	10.3 ± 0.22 μs	10.3 ± 0.16 μs	1
utils/count_constants/preserve_sharing	0.114 ± 0.0026 ms	0.112 ± 0.0025 ms	1.02
utils/count_depth/break_sharing	17.4 ± 0.39 μs	17.4 ± 0.37 μs	1
utils/count_nodes/break_sharing	10.2 ± 0.16 μs	10.2 ± 0.16 μs	1
utils/count_nodes/preserve_sharing	0.118 ± 0.003 ms	0.114 ± 0.0026 ms	1.03
utils/get_set_constants!/break_sharing	0.0533 ± 0.00083 ms	0.0529 ± 0.00077 ms	1.01
utils/get_set_constants!/preserve_sharing	0.324 ± 0.0072 ms	0.32 ± 0.0053 ms	1.01
utils/has_constants/break_sharing	4.52 ± 0.21 μs	4.49 ± 0.21 μs	1.01
utils/has_operators/break_sharing	1.93 ± 0.018 μs	1.94 ± 0.016 μs	0.998
utils/hash/break_sharing	30.1 ± 0.46 μs	30.2 ± 0.42 μs	0.996
utils/hash/preserve_sharing	0.133 ± 0.0029 ms	0.131 ± 0.0024 ms	1.01
utils/index_constants/break_sharing	28.2 ± 0.64 μs	27.7 ± 0.65 μs	1.02
utils/index_constants/preserve_sharing	0.127 ± 0.0031 ms	0.133 ± 0.0025 ms	0.955
utils/is_constant/break_sharing	4.35 ± 0.21 μs	4.36 ± 0.21 μs	0.997
utils/simplify_tree/break_sharing	0.248 ± 0.02 ms	0.17 ± 0.015 ms	1.46
utils/simplify_tree/preserve_sharing	0.377 ± 0.023 ms	0.286 ± 0.016 ms	1.32
utils/string_tree/break_sharing	0.579 ± 0.015 ms	0.569 ± 0.016 ms	1.02
utils/string_tree/preserve_sharing	0.721 ± 0.022 ms	0.706 ± 0.017 ms	1.02
time_to_load	0.676 ± 0.014 s	0.681 ± 0.0078 s	0.993