> > If I do that, would it be viable to offer subr-x on GNU Elpa as well?
>
> Like we do for seq.el? 

Yes, precisely.

> I don't like this duplication very much, but as
> long as they are 100% identical, I think it's OK.

I'm never a fan of redundancy, but in this case I think it's worth the evil.
But it would certainly be nice to have a way to link the source of an Elpa package to its built-in counterpart.