Hi Chris,
re: "I think we have different ideas of how pointing and tracking works", I think we're just getting sideways due to inconsistency in how folks use the terms "pointing model" and "mount model". We've gotten a bit off-track to my original post, but hopefully we can quickly get back on track. Let me try again to convince you (off-topic) on pointing's relation to tracking error, then I'll come back to summarize what I see as the on-topic issue. I'll refrain for a moment from using the term pointing model, and just use "sky model" to describe a model that doesn't know about mount imperfections.
Here's my assertion: A perfect sky model will command a perfect mount to a position that results in zero errors for acquisition AND tracking. Conversely, a perfect sky model used with an imperfect mount will result in positional errors during acquisition AND tracking. All mounts are imperfect. One might think that in this imperfect case, a sync command to fix acquisition would end the story. However, while the tracking start position would be (almost) corrected by a sync, the mount's imperfections remain, conspiring against tracking too (more slowly, but just as surely). The imperfect mount is just not able to move as perfectly as commanded by the sky model to guide any better than it can acquire due to those imperfections. The tracking error is smaller in magnitude because the size of each time interval's move is smaller, but it's there and it would accumulate (to the same magnitude as an acquisition error) if not corrected by guiding. This is why programs like TPoint and PointXP exist. They model a mount's underlying imperfections and pre-adjust commanded positions to account for those imperfections so that the mount can (almost) achieve what is desired. Folks tend to call the result of those programs "pointing models", but that's not quite consistent with what other people talk about when they use the same term to describe simple/perfect sky pointing modeling (i.e. without knowledge of mount imperfections). For the trivia minded, did you know that TPoint can create pointing models from just guiding residuals? They produce the same model as an acquisition error approach (assuming the same sky coverage in each approach). Hence my argument that acquisition AND tracking errors are caused from the same problem (imperfect mounts).
Back to topic: I started this thread when I saw Ekos' "Mount Model", in the align tab, with some metrics and concepts I recognize from TPoint and PointXP. I then wondered if I could use the "Mount Model Tool" to build and employ a pointing model as TPoint users would use the term...mount imperfections included. The caveat expressed about the mount's pointing model needing to use the sync left me unsure about my mount. I also didn't see any classical pointing error modeling terms I expected to see in the Ekos tool. Hence the posted question. The Celestron CGX-L mount manual makes no reference to any mount model error terms saved in the HC (excepting PEC). The Celestron "Pointing Model" (term loosely used) exclusive of CPWI software, appears to be just a simple sky model (lat/lon/time basis; no mount modeling terms excepting PEC). For users of CPWI software, there IS a capability to capture acquisition errors and create a mount model along the lines of what TPoint users would be familiar with, but the model runs internal to CPWI software. The HC and CPWI software have likely implemented very different pointing model fidelities. I hazard to guess that CPWI mount modeling exists precisely because the HC and mount don't have similar abilities. If true, that's unfortunate as many older tools (e.g. Argo Navis) implemented this long ago.
My best guess at present is that the Celestron HC has a very limited mount model (not as TPoint users would think of, only a simple sky model). Another guess is that Ekos align mount model has ability to measure and display acquisition errors that COULD be used to build a higher fidelity mount model, but may not actually do so in a way that can actually change mount performance. Finally, although CPWI does have high fidelity mount modeling capability familiar to TPoint users, I don't wish to use it. I'm trying to completely avoid CPWI in favor of KStars/Ekos. If Ekos can't build and effect a high fidelity mount model, I'll probably do a one-off exercise using CPWI and then assess how much improvement might be achieved. It's academic really; Guiding will work fine, even if it could work less hard with a true mount model in effect. It would only be interesting to know the magnitudes and types of mount errors seen in any modeling exercise.
Finally, for those interested in the Ekos capability who might not use it or remember it, see this post:
indilib.org/forum/ekos/2002-exciting-new...delling.html?start=0