Crowded node clusters?

[BlinksTale]

One of the big appeals of using Code Connections was developing a better understanding of how modular our code was at a quick glance. It's been great for highlighting that, tangled code, and generally a sense of what regions of our repo exist - but two different issues have made the dependency graph harder to read. Maybe these should be two separate issues instead of one?

1. What I would call "continents" (or islands?) of code on the map are often placing nodes very close to nodes from other continents/islands - even though there are no connections between those groups/continents/islands of code. I can share a screenshot of this later, but it means that at a glance it often looks like large sections of code are associated with each other even when there are zero dependencies.

2. Dead end or isolated branches of code often have their connections between nodes overlap with other branches of code, even though they have no association or dependencies. It would be great if a section of code (or group of classes rather) could be identified as self contained, and then really shoot off perpendicular to the rest of the code some great distance to represent that it is isolated. This would help for onboarding new programmers and showing them what classes did/didn't touch each other more quickly.

These are all "nice to have" features, but would make a major positive impact - so I figured that would be worth mentioning! I'm interested in these two changes enough that I might attempt them on my own even and do a PR if you don't first. :) But it would be great to have your insight on these too. Thank you!