Is there measurement that correlates with cohesive/pinpoint imaging?

Certainly you are familiar with the two slit experiment, just model a mid and a tweeter on a large baffle and let it rip for a few cycles.

as for precision matching of drivers and pairs, that is a hallmark of Vandersteen for 20 plus some years now, I know they all go in the chamber for a filter tweak as well - the 7 are matched to .1 db as I recall
- the matched pair are delivered with the actual plots. Obviously an unsymmetrical room can present challenges, as Eric suggests some treatments might be appropriate and of course there are fans of DSP.
Of course this fanboy has other speakers that have positives as well: Quad ESL 63, Thiel 2.3+, Dynaco A-25, some Totems.... some tiny baffle Sonus Faber that throw a super big 2D image ....

rather than focus on hyper flat frequency response maybe measure the RT-60 of your room.....

measure and listen but always enjoy

And the really excellent designers use science and listening and a deep love of music to move the art forward- stereo is an illusion after all. There are some excellent recordings utilizing research and products coming out of work on head transfer function - Macy Gray and Amber Rubarth come to mind - both excellent and hyper frustrating at same time. I will post LP numbers if desired....

I had a pair of Snell back in the day.... and Thiel, ESL63, Vandersteen..... I still have them, except....for the Snell. I certainly recognize the genius of Peter and they were quite good....

a curved baffle helps but I don’t think it’s SOTA any more....

Time and Phase. Study how the highly evolved ear brain function to localize food and threat sounds and you will understand it is small timing differences. Add in low diffraction because a large baffle destroys time information while functioning as a mechanical averaging machine ( frequency response )
read what JA has to say about Vandersteen “imaging champs”.... 
I will say if your references are studio multi track efforts you might be leaving a bit on the table relative to expectations, especially image depth

enjoy your speakers ! And the music