Actually, I'm pretty sure you can't, so I guess my question is, once you DO edit the clip map on a clip, how do you apply those changes to all the tracks in a project? I want to quantize a multi-track to the grid, but SONAR's initial guess was way off. Now that I've fixed the problems on the best track for reference how do I replicate?

Once you've edited the clip map you can extract a timing grid from it by selecting "Set project from clip" at your chosen resolution. "Clip" sets the average tempo, "measure" will provide a tempo change at every measure, and "beat" a tempo change at every beat.
 
Editing the clip map doesn't actually change the audio in any way, it just makes sure that the audio transients of a clip line up with what Sonar thinks is the correct beat, thereby making the process of tempo extraction more accurate.
 
Once you've got the tempo map extracted and applied to the project you can then select all clips and have them follow the project.
 
I always have more success if I set the project tempo to very roughly what the actual tempo is to start with.

Yah, but I don't want to tempo map the project, I want to quantize them to a hard grid (so I can do other tempo related functions later). But most of the clips have long gaps of silence since they're multitrack parts that don't appear through the whole song. I can get SONAR's tempo close, but since the performance was recorded without a click it's not exact, and I don't know how to apply the changes to the other tracks. 
 
As an experiment, I tried to do this in Reaper and Live last night. I didn't expect to succeed fully but I did confirm that it's not easy-peasey in other DAWs...which I suppose is comforting in some weird way. 
 
Surprised there's not a good example of how to pull this off, it seems like a pretty common problem if you get stems from a live band that you want to do a dance remix for. First you want to quantize the whole thing to a grid and then you can sequence and manipulate tempo globally (like adding accelerandos and whatnot). 

Even if you could just copy/paste transient markers to other tracks... that would make this easier too. Or create a "project transient" (which I reckon is essentially a tempo marker)
 

I found this interesting thread last night. I'm going to try and use this method (maybe some variation on it) to see what happens:
 
http://forum.cakewalk.com/Audio-Snap-2-multi-track-drums-quantize-in-phase-m1921825.aspx
 

...wicked
Yah, but I don't want to tempo map the project, I want to quantize them to a hard grid (so I can do other tempo related functions later). But most of the clips have long gaps of silence since they're multitrack parts that don't appear through the whole song. I can get SONAR's tempo close, but since the performance was recorded without a click it's not exact, and I don't know how to apply the changes to the other tracks. 
 
As an experiment, I tried to do this in Reaper and Live last night. I didn't expect to succeed fully but I did confirm that it's not easy-peasey in other DAWs...which I suppose is comforting in some weird way. 
 
Surprised there's not a good example of how to pull this off, it seems like a pretty common problem if you get stems from a live band that you want to do a dance remix for. First you want to quantize the whole thing to a grid and then you can sequence and manipulate tempo globally (like adding accelerandos and whatnot). 

 
That's what I mean by extracting a tempo map. That produces the hard grid that you can quantize too later. I use that quite a lot for freely recorded parts.
 
Assuming that the tracks are in time with each other there's only a need to extract the tempo from one of them.
 
 

...wicked
Even if you could just copy/paste transient markers to other tracks... that would make this easier too. Or create a "project transient" (which I reckon is essentially a tempo marker)
 

You can. The function to do that is the "Merger and lock markers" command found in the right click menu. Select all the tracks you want to use and use that command. Use the threshold setting to hide all markers on overhead tracks if working with multi mic'd drum parts.

Uh, maybe I'm not reading it right.
 
If you set the project to follow the clip, SONAR just inserts tempo changes so that it's measure boundaries line up with the existing audio. I'm talking about defining the measure boundaries and than having SONAR stretch the audio so that it's mapped measure boundaries line up to where they would at the project tempo in time. So the grid would be, er, all the measures would be the same real-time length as if it had been quantized (which it has), but so that ALL the tracks are quantized the same way, with the measure break info coming from the one track's clip map. (does this make any sense? :-)
 
As for your second answer, I'll go look that up!
 

Aha, now I"m getting somewhere. Thanks for the tip about the pool options. I managed to add the guide track transients to the pool and apply those to all the other clips. Now I can quantize them and success! There's just a few places where it sounds funky so I"m going to play with it a bit. But hey I"m off and running!
 

What you want to do can get very complicated and may take multiple attempts to get right, but you're on the right track (no pun intended). Basically its a three-step process (four, including the final bouncing of clips to render the stretching):
 
1. Set the project timeline to match the audio of your reference track using Set Project From Clip as Karl described. You have to do this so that SONAR knows where the timing irregularities are that you want to eliminate.
 
2. Select all the clips, including the reference track, choose the Autostretch option from the dropdown to the right of Clip Follows Project, and then Enable that on all tracks.
 
3. Now you can alter or delete any or all of the tempos in the timeline and SONAR will stretch/compress all the clips equally to conform to the new tempos or to a single, initial tempo if that's what you want.
 
The level of precision you use in the first step will determine how tightly everything is "quantized". In theory, you could bounce the stretching, and then do additional quantizing of individual tracks at higher resolutions as appropriate, but I think you're likely to get bad artifacts from stretching more than once. As it is, you may not like the results you get. The less you mess with the tracks to get them where you want, the better.  A lot depends on the nature of the content. Also, once all your tempo changes are done, it's important to select each clip individually and Bounce to Clip(s) to render the stretching permanent using the most appropriate "Offline" algorithm in the Audiosnap Palette for that clip's content.
 
All that said, here's the bottom line: It's going to be a ton of work, and I don't think you're going to like what you get. It were me, I'd stop after step 1. That will give you a grid that you can use to add or replace sounds while maintaining the original ebb and flow of tempo. And if an individual part has wonky timing, you can quantize specific parts of specific tracks to tighten them up against other parts without changing the overall tempo variation.
 
EDIT: Oops. Stepped away form a half-finished post for a bit, and see you've come up with an alternate strategy in the mean time.

No no, it's great. I'm pleased this is possible, but confounded how arcane it was to find all the info. Between the pallette, the right-click menu for a clip, and the clip itself it's all there but there's little clues as to workflow OR that they're relevant. But hey, it's doable!
 
I would otherwise just keep the natural tempo variations as I prefer those to be in there, but the song itself has a second half that is a deliberate tempo change. in the original it appears as a sudden change, and I want to turn it into a gradual change. So, being able to continue to muck with the tempo post-quantizing is imperative. Granted, it's a strange example but I COULD see dance remixers wanting to change the tempo of stems for remixes, so I still think it's common enough of a function to want to achieve.
 
I'll keep playing with it. Maybe write up a recipe.