Now, I split the entire track at every measure, rename the clips for organization and, in an attempt to render the AudioSnap edits into each clip, I "Convert to Clip" for each slice.
If by "Convert to Clip" you mean "Bounce to Clip(s)", this is the correct way to render AS modifications permanent. But once done, SONAR will re-detect transients in the new clips, and it is likely to make the same detection errors it did the first time, making it look like things are fouled up again. You can ignore the results of re-detection. It should play correctly (within the limits of how well the stretching algorithms work, which may not be that great, depending on the material).
However, I would suggest you avoid all the splitting an Audiosnapped clip without first rendering it. Initially, you are creating a slip-edited copy of the whole AS clip for every slice that contains AS transient info for every marker in the original full-length clip. This is what slows down the saving and loading of the project before all the changes are bounced down. The audio file is not duplicated, but all the transient marker info is.
If you really need to split the clips up, you should do it
after bouncing the Audiosnapped clip down as a whole. And in any case, rather than freezing or bouncing the track in place, you should bounce the whole Audiosnapped track to a
new track. This way, you have the best of both worlds - a bounced track with its own, destructively edited audio file that preserves your initial edits regardless of what might happen to the Audiosnapped clip, plus access to the original track with AS edits if you decide you need to go back and fix something later.
To be 100% immune from "Audiosnap shinanigans" (other than possible losing time), you might even want to start with a bounced copy of the original track to do the Audiosnapping on, and Archive the original.