In rare cases, next_random_seed() fails

[jemus42]

..or at least I am reasonably certain it does, because I found that the implementation here
https://github.com/DavisVaughan/furrr/blob/aa124bc06d9fc3e18ea047cb90d270ebeb8d9698/R/seed.R#L66-L73

uses the same validity check stopifnot(!any(seed_next != seed)) used in future and future.apply, see issue here:
futureverse/future.apply#122

The short version is that the underlying representation of NA values leads to a rare bug where the next seed drawn is actually NA rather than an integer, and the comparison fails.

[HenrikBengtsson]

Thanks. I've created PR #272 fixing this.