Pass RNG as first argument

[devmotion]

It seems the convention in Julia is to pass the random number generator as first argument, see e.g. the documentation of rand, rand!, and randn!. I changed the definitions of count_rand, ad_rand, and pois_rand accordingly.

Moreover, when I updated the README I realized that not only the benchmarks do not consider some already existing implementations of the algorithms in this package in Distributions but also compare the performance of different random number generators - n_dist uses Rmath's internal random number generator whereas the other functions (n_count, n_ad, and n_pois) use rng = Xorshifts.Xoroshiro128Plus(). I'm not sure whether this could explain the differences seen in the benchmarks but at least it does not seem to be completely fair.

[codecov]

Codecov Report

Merging #4 into master will decrease coverage by 0.73%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master       #4      +/-   ##
==========================================
- Coverage    55.4%   54.66%   -0.74%     
==========================================
  Files           1        1              
  Lines          74       75       +1     
==========================================
  Hits           41       41              
- Misses         33       34       +1

Impacted Files	Coverage Δ
src/PoissonRandom.jl	54.66% <100%> (-0.74%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 82d88c6...96d39e7. Read the comment docs.

[ChrisRackauckas]

Oops, thanks. Can you make sure an update gets to DiffEqJump as well?

[devmotion]

Sure, I also didn't consider the fact that it's a breaking change... I checked METADATA and it seems that DiffEqJump is the only registered package that depends on PoissonRandom, so I guess it's not necessary to add a deprecation warning and we can just upper bound existing releases of DiffEqJump, release a new version of PoissonRandom, change the order of the arguments in DiffEqJump, and then release a new version of DiffEqJump.