Correct, more often than not if you do a trajectory trace, our EML's derive from the southwestern U.S. I mean you can also get a bit more technical and say more the southern-tier of the Inter-mountain West region.
You are correct, these are characterized by very dry mid-levels with sharp warming usually around 800mb-700mb then a rapid decrease of temperature with height above that (hence the steep lapse rate. And yes, typically they will traverse the crest of strong mid-level ridging with a westerly-to-northwesterly flow in the mid-levels.
As for why they aren't more common the biggest factors really is the Gulf of Mexico and convective overturning. East of the Mississippi River (perhaps maybe even a bit farther west than this) the Gulf of Mexico is a major contributor to the climate of this region. For example, if you were to look at a national annual rainfall map you would see a significant gradient right around the Mississippi River. Warm/moist air from the Gulf of Mexico transporting northward (especially when you have a deep southerly flow) disrupts the structure of the EML so the capping associated with that warm layer is weakened which promote more mixing.
In terms of convection, when you have widespread convective events you get outflow boundaries, convective overturning, latent heat release, turbulence, etc...all of these promote mixing which degrades the integrity of the EML.
Michael Ekster (I also think radarman was an author as well) wrote a paper about 10-years ago regarding EML advection into the Northeast. It seems the best synoptic configuration for an EML to maintain it's integrity into our region was to have a strong 700mb height anomaly in the southeast U.S. (with the strongest anomalies I think in the TN Valley region) with a trough digging into the western U.S. This flow seemed to promote the northward transport of EML air farther north into the northern Plains...even into Canada where it is farther removed from the Gulf of Mexico, thus limiting potential influences from this moisture source. Also, this seemed to be yield the maintaining of the cap associated with the EML preventing mixing.
There was also something else of interest too...actually will have to go back and refresh my memory on this but there was something weird about the trajectory course of previous EML events studied where the trajectory did a loop...at the time it wasn't sure what this was and why.