I am seeing that for point 4 the GCP elevation value was 419. This appears to be close to the value in the DEM. It was 429 or so in the original.
Point 7's elevation value was 399 in your GCP file which appears to be reflected in the adjusted version. I was 408 or so previously.
Keep in mind that the only non-linear adjustments occur in the X and Y directions. The vertical shape of your model is not adjusted based on your GCP inputs. Only a linear adjustment/offset is applied. The error is minimized for all GCPs so the individual values may not be right on.
Basically your model isn't going to get any more accurate in a relative sense using this process. If your model is bad to start with it isn't going to be accurate everywhere.
You had pretty minimal overlap in certain sections which does not help the model's accuracy. If you are going to care about the accuracy of the elevation numbers your entire Overlap Report needs to be in the dark blue zone which corresponds to using 80/80 overlap and taking off from the highest point you are going to be covering.