I don't have nearly such large flows as that.
Instead I put any self-contained processes on a separate editor tab and use link calls to connect to it.
Examples include database insert/select, email or SMS notifications etc.
No doubt other people prefer subflows for such tasks.
I also put the nodes which perform a defined function into a group together.
I try to line up nodes vertically and horizontally.
Within the group I use junctions to prevent wires crossing and passing beneath nodes.
Connections from one group to another mostly use link calls.
Like this it's easier for me to identify the flow and nodes to update when I need to modify a program.
Here's an example
Flows with widely spaced nodes, curvy wires and a lot of link nodes give me a headache:
