Clean up besselj(nu,x) and add docs

[heltonmc]

I've added more docs to besselj and other internal methods.

I've also cleaned up the besselj implementation to be more consistent with bessely while removing several branches due to using the Hankel expansion instead of using either forward or backward recurrence. I think this helps with branch prediction and speeding up the overall routine. Accuracy is unchanged. I've also removed the loggamma implementation from SpecialFunctions.jl so this function is entirely in Julia now.

julia> @benchmark Bessels.besselj(y, x) setup=(x=rand()*300+20.0; y=rand()*300)
BenchmarkTools.Trial: 10000 samples with 963 evaluations.
 Range (min … max):   84.805 ns … 580.607 ns  ┊ GC (min … max): 0.00% … 0.00%
 Time  (median):     146.028 ns               ┊ GC (median):    0.00%
 Time  (mean ± σ):   178.862 ns ± 115.707 ns  ┊ GC (mean ± σ):  0.00% ± 0.00%

  █ ▆▅     ▆▇   ▅▄                ▂▂▁▂▂▂▂▁▂▂▂▂▂▂▂▁▁▂▂▁▂▁▁▁▁▁▁▁▁ ▂
  █▆██▇▅▄▄▃██▇▆▄██▆▄▃▄▃▁▁▁▁▁▁▁▁▁▁▁█████████████████████████████ █
  84.8 ns       Histogram: log(frequency) by time        475 ns <

 Memory estimate: 0 bytes, allocs estimate: 0.

julia> @benchmark SpecialFunctions.besselj(y, x) setup=(x=rand()*300+20.0; y=rand()*300)
BenchmarkTools.Trial: 4171 samples with 535 evaluations.
 Range (min … max):  207.865 ns … 7.374 μs  ┊ GC (min … max): 0.00% … 0.00%
 Time  (median):       1.464 μs             ┊ GC (median):    0.00%
 Time  (mean ± σ):     2.237 μs ± 1.453 μs  ┊ GC (mean ± σ):  0.00% ± 0.00%

  ▄▅ ▂   ▁▄█▆▅▁▁    ▄▆▅▁▂▂▂▁  ▃▃▂     ▃▄▄▃▁▁                  ▂
  ████▃▃▅████████▃▃▅█████████▇███▇▃▄▁▇███████▅▆▆▅▃▁▁▁▃▄▃▁▇███ █
  208 ns       Histogram: log(frequency) by time      6.83 μs <

 Memory estimate: 16 bytes, allocs estimate: 1.

[codecov-commenter]

Codecov Report

Merging #28 (4fd03bd) into master (a09b94d) will decrease coverage by 0.02%.
The diff coverage is 93.75%.

@@            Coverage Diff             @@
##           master      #28      +/-   ##
==========================================
- Coverage   99.06%   99.03%   -0.03%     
==========================================
  Files          14       14              
  Lines        1172     1145      -27     
==========================================
- Hits         1161     1134      -27     
  Misses         11       11              

Impacted Files	Coverage Δ
src/asymptotics.jl	98.29% <ø> (+0.55%)	⬆️
src/bessely.jl	100.00% <ø> (ø)
src/besselj.jl	99.25% <92.85%> (-0.75%)	⬇️
src/U_polynomials.jl	99.28% <100.00%> (ø)

📣 Codecov can now indicate which changes are the most critical in Pull Requests. Learn more

[heltonmc]

I'm going to merge this before #29 and deal with any conflicts.