The data collection and overlap all look good. Usually we get to within a pixel or less in the alignment but the vertical numbers depend on the accuracy of the model and the accuracy of the collected points.
Since we don't know how good or bad the elevation accuracy is and have no required for any specific number all we can do is minimize the vertical error among all the points using the the shape of the original model that was calculated. Essentially, we don't change the shape of anything for the realignment other than to match the GPS locations in latitude and longitude. This is why we don't offer any of the 3D outputs on the reprocessing. Only that image based products are realigned. You will likely want to use client software on your computer if you are really trying to get a solid RTK accurate result. Things like that take a lot of touching and knowledge to do properly.
For volume measurements this almost never matters since those are relative measurements. The general shape of the model is accurate in relation to itself over short spans. If you are looking for absolute accuracy you will likely need to get good at doing a lot of this by hand/manually.