How to practically implement Automated Factory Building

[kwiz]

I propose that a mobile, track-based unit capable accepting a macro be able to run along this track and lay down a set of blueprints in the order and orientation that you input them, and start over from the first blueprint.

How this would work:
Say that this is the track with the blueprint layer at the start of it: O>=========== and the > shows the orientation of the blueprint layer.
Now you take a blueprint, rotate it to the orientation that you want, and set it ON TOP OF THE BLUEPRINT LAYER, at which point it accepts the blueprint. The blueprint needs to have track for the layer to run on, and another blueprint layer inside of the blueprint to show the blueprint layer that is doing the laying where to travel to after lying the current blueprint. So here is our new setup: ==FACTORY==O>===========
In input mode (where looping is disabled) the blueprint layer would then stop at the location of where the blueprint layer in the blueprint was, and await your next blueprint (centered upon its new position). This would be blueprint 2, to be laid after laying blueprint 1. You could then add a blueprint 3, 4, 5...until you felt that you wanted to repeat the sequence. At this point, you would switch the mode of your blueprint layer, and it would begin to execute the sequence, repeating after reaching the last blueprint. This would allow a factory to expand in a dumb, linear way (following whatever train tracks you put in the blueprints) automatically, with construction robots expanding your factory for you ad nauseum. Lastly, the blueprint laying process would automatically deconstruct anything not in the blueprint (IE factories, forests).

Considerations:
Do we want an inborn roboport on our blueprint layer, with the layer proceeding only after completing its current blueprint (and also making the max use of a layer-centered blueprint) or do we want the layer to only lay blueprint and require a roboport in the design to actually construct it? The first option is probably easier to use and easier to code. The other way that the layer could behave before moving on to the next blueprint is to always have the blueprints it lays adjacent to each other, so it would lay blueprint #2 when that blueprints edge met the edge of blueprint 1 (and obviously only really be able to do that when there was track to allow it to move forward. The problem with this is that you would have to have the layer discount railway tracks, and really go by the edge of non-railway components in the blueprint.
This obviously isn't going to be able to take advantage of mineral deposits or to take care of biter nests in the way, but that is okay. Ideally, the player shouldn't be able to leave the game on overnight and come back to a 2 GW factory in the morning. If I'm alone in this opinion, a railway gun https://en.wikipedia.org/wiki/Railway_g ... 27627u.jpg with a small angle could be a separate technology to research and be attachable to the blueprint layer to bombard the biter bases from long range. If you took this route, however, you would need to consider how to leave ample time to eradicate the biters, and probably write in code to detect biter proximity and stop the system until it was resolved.
As far as Technology goes it obviously has robotics and railway dependencies. Multi-tier technology of this might include a larger macro limit, possibly the ability to record blueprint rotation, or maybe the ability to attach the aforementioned railway gun to deal with biter nests in the way.
This may have good synergy with terraforming mods.