wanted: selective (aka filtered) model merging

it would be really, really nice if there was some kind of advanced node, that could do selected, filtered merging.

Instead of straight global percentages, identify the weights from the merge candidate model that are particularly associated with one or more concepts (ie: a prompt) and merge THOSE nodes... but ignore non-associated weights.

By "associated", I mean "within distance X of the concept". So the node would have two tunables: a prompt, and a distancevalue.

IDEALLY, it could have an (optional?) feature to calculate furthest-away distance in the model, so that the user could have a better idea of practical range to use for distance.