Doc: Proper attribution

[PallHaraldsson]

[skip ci]

[PallHaraldsson]

FYI: I contacted Prof. Vigna, and he wrote back (I'm skipping most of what I wrote and what he wrote back):

> It's taken a while, and I only did advocate and look into options, not just your algorithm, others did the actual work, Jeff and others, which alone seems would have been easy to add.

Nice!

[..]

> At least you might know if Julia is NOT the first (as a default or option, your work was available before in a package, most do not go looking). I would find that interesting too.

GNU Fortran uses xoshiro256**.

[..]

> In the code:
> /* This is xoshiro256++ 1.0, used for tasklocal random number generation in julia.
> This implementation is intended for embedders and internal use by the runtime, and is
> based on the reference implementation on http://prng.di.unimi.it
>
> Credits go to Sebastiano Vigna for coming up with this PRNG.     
>
> There is a pure julia implementation in stdlib that tends to be faster when used from     
> within julia, due to inlining and more agressive architecture-specific optimizations.

Nice. If you can remove the "Credits" line and quote the paper, better, as I'm not the only designer--there's also David. Like:

---
/* This is xoshiro256++ 1.0, used for tasklocal random number generation in julia.     
This implementation is intended for embedders and internal use by the runtime, and is     
based on the reference implementation on http://prng.di.unimi.it. xoshiro256++ has been
described by David Blackman and Sebastiano Vigna in "Scrambled Linear Pseudorandom
Number Generators", ACM Trans. Math. Softw., 2021.
---

Ciao,

                                        seba

I'll write back to him pointing to this PR, and I see now he suggested dropping the Credit line. Seems strange, I was copying his text and only "described" seemed off. Please suggest changes if needed, also to the other file. I can fix, I see a typo anyway still in there: "agressive".