Good point: setting constraints can be misleading unless you "know" the constraint is true.
As an exercise I tried using SR on the 6 observations (without any attempt to improve the initial automatic solution).
The clones were displayed in planetarium software with date set to to about 36 hrs after the observations. The clones showed up as a tight elliptical cluster about 1 arcmin long plus a sparse "fan-tail" stretching another 4-5 arcminutes from the main cluster. The "true" position would be about 1.5 arcmins from the centre of the cluster part way into the tail.
Looking one week after the date of the observations the main cluster stayed in an ellipse 4 arcmins long, the fan-tail had spread to 1 degree and the "true" position was about 8 arcmins from the cluster into the tail.
For an object like this, the SR approach seems to provide good guidance for follow-up observations for several days into the future. I suspect things are not so simple with a high-score NEOCP. Must have a go with one.