This should fix #16 . @lopezpaz can you do a code-review? In particular, I can't think of any edge-cases / interactions with sweep etc. that this breaks, but this has historically been a tricky part of the codebase.
Should we add algorithm to the arguments of seed_hashhere? Otherwise if two algorithms have a parameter with the same name, their random values would be tied, even if they correspond to different things or have different scales. The rest looks good to me!
This should fix #16 . @lopezpaz can you do a code-review? In particular, I can't think of any edge-cases / interactions with sweep etc. that this breaks, but this has historically been a tricky part of the codebase.