It works on my MBP 13" 2012 as well, so it doesn't depend on whether or not having a dGPU.
I have to assume, unless you clarify otherwise, that you have a NON-retina 13" MBP from 2012. Your experience is an interesting data point; however, I feel compelled to point out that we are talking about rMBP in this thread, which I don't think is insignificant (especially since that's what the OP was asking about, after all). Still, the 2012 13" cMBP had Intel HD 4000 graphics, same as the late-2012 13" rMBP, so you might expect results to be similar.
The difference may come in the EDID from the LCD panels themselves. Here's a question for you: how long is the EDID that is being returned to you? Run the same command, but without the "| xxd -p -r | strings -6" at the end, and then (ignoring the > at the tail end) count the length of the string that is returned to you. Is the hexadecimal number that you see there 256 digits (128 bytes) long, or is it 512 digits (256 bytes) long?
If it is 256 digits long, then your non-Retina panel has a standard-length version 1.3 EDID. I see the same thing on my MacBook Air. For every retina panel that I've seen in the wild, however, there is a 512 digit long EDID, so a 128-byte-long EDID with an additional 128-byte-long E-EDID appended to the end.
The actual panel details (including manufacturer-assigned model #) are only included in the E-EDID on these retina panels. If you look at the results from the grep (now removing the "| sed" bit and everything past it), on a Retina machine with discrete graphics, you will see that it picked out two results from the IOService tree matching the string "EDID": one called EDID and one called IODisplayEDID. The value of the EDID key is twice the length of the IODisplayEDID one (on a Retina machine with discrete graphics), which matches the former exactly except that it is truncated after the 256th digit. On a Retina without discrete graphics, only IODisplayEDID shows up, and it can be surmised that it is also truncated (but we don't have the whole EDID to compare it to on those machines, obviously). The EDID property exists under the NVDA,Display-A@0 branch (NVDA == nVidia, clearly) and IODisplayEDID under display0\AppleBacklightDisplay, which is quite a bit more generic sounding. The former is pretty evidently being generated by the nVidia driver, which of course isn't being used on non-discrete-graphics models.
On my 2010 MacBook Air that has the nVidia 320M, I also see both EDID and IODisplayEDID. On this machine, both strings match exactly, and both are only 256 digits long (128-bytes in length).
So my hypothesis is that on a machine with nVidia graphics, the nVidia graphics drivers are able to read and pass on the entirety of the EDID, but on other machines with Intel graphics, we are only being passed the first 128-bytes-worth of the EDID. On non-Retina machines, the panels that are being used only contain a 128-byte-long EDID, which can be successfully interrogated even on non-discrete-graphics models. But on Retina machines, the panels being used have 256-byte-long EDIDs and the contents of the panel's EDID cannot be retrieved past the 128th byte on models without nVidia graphics. Thus, at the present time, there is no way to retrieve complete panel details on *RETINA* MacBook Pros that do not have nVidia discrete graphics.
-- Nathan