Step size initialization uses GLOBAL RNG instead of reproducible RNG from caller

[getzdan]

AdvancedHMC.jl/src/trajectory.jl

Line 770 in 6a55a3f

return find_good_stepsize(GLOBAL_RNG, h, θ; max_n_iters=max_n_iters)

This caused an actual reproducibility problem, see Discourse thread:
https://discourse.julialang.org/t/stablerng-in-turing-is-not-producing-reproducible-output/92032

To get RNG from caller, it might need to be perculated down from callers, with a few other changes needed.

[devmotion]

AFAICT the issue was fixed by TuringLang/Turing.jl#1924. As for other RNG-dependent methods, users are free to use the convenience function that uses the default RNG. All internal uses of find_good_stepsize in AdvancedHMC and Turing use the version with explicit RNG.