That would mean gutting almost everything out into a separate plugin. Since we're already struggling to keep up with the inflow of new minecraft features, there's not enough manpower for that. Also, it would most definitely mean some loss of speed, which is not what we want.
The general direction in which we want to be moving is to be the server software of choice for hosted servers - you rent a (virtual) server, compile MCServer there and run it with as much speed and as little resources as possible. Let's leave the extreme extensibility and pluggability to Java, it's better suited for that task asnyway.
Of course this doesn't mean that you can use MCServer source and base a new server off of it; feel free to do that, that's what opensource is mostly about.
The general direction in which we want to be moving is to be the server software of choice for hosted servers - you rent a (virtual) server, compile MCServer there and run it with as much speed and as little resources as possible. Let's leave the extreme extensibility and pluggability to Java, it's better suited for that task asnyway.
Of course this doesn't mean that you can use MCServer source and base a new server off of it; feel free to do that, that's what opensource is mostly about.