Improve performance of `JohnsonSU`

[devmotion]

master

julia> using Distributions, Chairmarks, Random

julia> @be randn logpdf($(JohnsonSU(10, 10, 1, 3)), _)
Benchmark: 3581 samples with 873 evaluations
 min    28.016 ns
 median 29.353 ns
 mean   29.809 ns
 max    77.606 ns

julia> @be randn pdf($(JohnsonSU(10, 10, 1, 3)), _)
Benchmark: 1634 samples with 1235 evaluations
 min    23.111 ns
 median 23.245 ns
 mean   23.672 ns
 max    44.500 ns

julia> @be randn(Float32) logpdf($(JohnsonSU(10.0f0, 10.0f0, 1.0f0, 3.0f0)), _)
Benchmark: 3243 samples with 1173 evaluations
 min    24.474 ns
 median 24.546 ns
 mean   25.004 ns
 max    51.648 ns

julia> @be randn(Float32) pdf($(JohnsonSU(10.0f0, 10.0f0, 1.0f0, 3.0f0)), _)
Benchmark: 3319 samples with 1240 evaluations
 min    23.051 ns
 median 23.152 ns
 mean   23.333 ns
 max    38.273 ns

julia> @be rand!($(Random.default_rng()), $(JohnsonSU(10, 10, 1, 3)), $(Array{Float64}(undef, 100)))
Benchmark: 3271 samples with 33 evaluations
 min    843.424 ns
 median 862.364 ns
 mean   870.551 ns
 max    1.492 μs

This PR

julia> using Distributions, Chairmarks, Random

julia> @be randn logpdf($(JohnsonSU(10, 10, 1, 3)), _)
Benchmark: 4171 samples with 696 evaluations
 min    27.418 ns
 median 30.172 ns
 mean   31.702 ns
 max    161.579 ns

julia> @be randn pdf($(JohnsonSU(10, 10, 1, 3)), _)
Benchmark: 3907 samples with 871 evaluations
 min    22.436 ns
 median 22.675 ns
 mean   24.059 ns
 max    45.398 ns

julia> @be randn(Float32) logpdf($(JohnsonSU(10.0f0, 10.0f0, 1.0f0, 3.0f0)), _)
Benchmark: 4145 samples with 858 evaluations
 min    22.970 ns
 median 25.690 ns
 mean   26.697 ns
 max    69.882 ns

julia> @be randn(Float32) pdf($(JohnsonSU(10.0f0, 10.0f0, 1.0f0, 3.0f0)), _)
Benchmark: 2989 samples with 1428 evaluations
 min    19.608 ns
 median 21.943 ns
 mean   22.560 ns
 max    39.391 ns

julia> @be rand!($(Random.default_rng()), $(JohnsonSU(10, 10, 1, 3)), $(Array{Float64}(undef, 100)))
Benchmark: 4086 samples with 27 evaluations
 min    750.000 ns
 median 813.259 ns
 mean   845.845 ns
 max    1.546 μs

[codecov-commenter]

Codecov Report

:x: Patch coverage is 94.73684% with 1 line in your changes missing coverage. Please review. :white_check_mark: Project coverage is 86.30%. Comparing base (efff906) to head (08a653d). :warning: Report is 9 commits behind head on master.

Files with missing lines	Patch %	Lines
src/univariate/continuous/johnsonsu.jl	94.73%	1 Missing :warning:

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1976      +/-   ##
==========================================
+ Coverage   86.29%   86.30%   +0.01%     
==========================================
  Files         146      146              
  Lines        8791     8805      +14     
==========================================
+ Hits         7586     7599      +13     
- Misses       1205     1206       +1     

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

:rocket: New features to boost your workflow:

• :snowflake: Test Analytics: Detect flaky tests, report on failures, and find test suite problems.