I think with the current rMBP, they are saying: the MBA experiment worked well, if we give such a design very high end specs, it should please 99.99% of the high end users even without the capability to swap components.
The MBA experiment did not prove anything about the bulk of the Mac laptop buyers which is what is the critical issue. Not the upper 10% Mac laptop buyers.
The MBA exposed some flaws in Apple's minimalist initial design. One USB port and crippled I/O ( wifi only) is not enough. The current MBA has two USB 3.0 port (and two USB 2.0 ports immediately before) and Thunderbolt. The minimal at all costs is flawed.
Another example was when Apple prematurely nuked the FW ports off the immediate precusor to the MBP 13" design. Another bonehead move by Apple.
The MBA sales volume before the MBA 11" was relatively anemic. It regularly ranked below the Mac Pro in "top macs sold" on Apple's online store back when Apple let the Macs sales rankings be indirectly exposed. The lowest selling Mac model overall in terms of volume.
MBA is much higher now primarily because the 11" is the lowest price laptop Mac available. In that sense perhaps you're correct in that Apple may just "give" the MBA 13' the win not what but I doubt they would really want to gamble that way with the volume leader of the Mac laptop line-up.
The rMBP 13" isn't going to carry the water for "high volume Mac laptop" when it gets released. Its price is extremely likely to be much higher then either MBA or MBP 13".
I suspect that most MBA 13" seeking the "high price premium" 13" model will move up and select the rMBP 13" because they are the most similar. Those who are more price sensitive will buy a MBP 13" and hence drive volume sales.
It's quite possible, as you suggest, that Apple takes a wait and see approach and axes the one that is least popular. But I wouldn't expect so because I consider Apple to be a company of principles.
Yeah they have principles and one of those is selling what people want. This is a striaghtforward test they can conduct here. Lots of folks assume the answer is going to go the MBA 13". It may or may not.
I don't see it that way. Both the MBP and MBA have exactly the same dimensions, except for the thickness. If the rMBP is as thick as the MBA at its thickest point, it would be about twice the volume and therefore much closer to the MBP in feel.
"feel" isn't the primary issue.
Might as well fix them and improve in areas that most users do appreciate: weight, size, battery life.
That's the problem with a slimmed down rMBP 13" versus the MBA 13". It really doesn't put up a clear win on any of those. If there is a relatively small ( 0.5-0.8) weight difference and equal or better battery life with a much better processor ..... It loses. Its primary advantage is really just being cheaper.
The volume/size difference being overblown here. I think folks will trade off the taper for the better life and "twice as good" screen. The MBA taper is an extremely dubious sacrifice if that blocks you from a substantially better screen. A 07" rectangle is thin. Quite thin. It also wouldn't hurt to have two Thunderbolt ports also (instead of throwing that away too just for some "taper". Go back to 1 versus 2 USB socket evolution above for very similar reasons. )
Over time as the retina displays got cheaper the line-up could collapse some more.
In my very humble opinion, the days of user upgradable computers are as numbered as the days of user upgradable cars.
As long as Apple charges sky high mark-ups on memory and SSDs, this is a flawed analogy. It isn't so much folks want to crack open the computer and install these. it is much more so that they want better prices. The desire for better pricing doesn't go "old fashioned".
If Apple adjusted they BTO pricing to be more market driven, it wouldn't be an issue but until they do it is. The primary issue here is not "old" vs. "new" designs.
Over 2-3 more design iterations Apple may close out the classic MBP. I don't think prematurely killing them off is a winner for Apple. They can still pull substantial profits by iterating another iteration until get SoC Haswell (or follow on ) offerings that let them shrink the smaller rMBP designs and still get a discrete GPU in there.