bug#73431: Add `setf` support for `stream.el` in ELPA

[Stefan Monnier]

>> Maybe such examples would be better in a shortdoc.
> I dunno, but the example also tells something essential: that the
> mapping happens in a delayed way.  That, and the implications, might not
> be obvious.  Maybe it's not wrong to have that in a docstring - maybe
> even more explicit than currently?

AFAIK, `seq-map` and `seq-mapn` always return a sequence of the same
type (same type as the first sequence in the case of `seq-mapn`).
Maybe their doc should say so more explicitly.
That info would be sufficient to know that the `seq-map(n)` does its
job lazily when applied to a stream (it would be a clear misfeature to
return a stream and yet to build that stream eagerly).


        Stefan