If we combine the latter with the sensible features of NL2P, we could lose anything but it in the core.
Note that nl2p probably needs a complete rewrite to be useable as default, cause it really produces some mistakes right now (browser can handle it, but it don't validate). And at the moment i can't think of a way to correct these issues without reworking it more to a proper parser.
At topic: I was in the situation of not wanting to kill nl2p with textile and markdown, not using wysiwyg and being sick of using pure html all the time and the s9ymarkup-plugin was not powerful enough (i missed especially linking stuff). So i made liquid, so far:
*italic*, **bold**, [url linkname], [[imageurl]], [url title linkame], [url].
The one thing missing in my "daily" blogwork is creating lists.
I mention it because as far as i remember, i based it upon the code of s9ymarkup. Extending it was quite easy, not breaking nl2p is also easy if the other markup-plugin don't replace the \nl. So we really could try to continue from there, extending s9ymarkup (or liquid?) if necessary and combine it with nl2p (regarding it as part of the markup-tranformation).
It should be possible to offer this as a default-markupplugin without having to rely on the really powerful and complex textile/markdown-plugins (which i also didn't want to use because i couldn't fix a bug in there), and we could try to prepare the editor to work with that (configure the buttons accordingly).
But of course, it could be a valid alternative to don't change anything in here, or setting textile/markdown as a default. I saw quite some blogengines going the latter route.