Cleanup Code (dear god, its terrible)

[SteveMacenski]

• [ ] pull out visualization stuff into separate class
• [ ] call slam_toolbox library in application ROS level, too much is happening inline.
• [ ] perhaps separate sync and async nodes
• [x] node bringup with a file to work with rather than a service
• [x] parameterize live slam w/ and w/o decay
• [x] messy node with everything but flexible, or changing between them cleanly

[SteveMacenski]

200k synchronous (processing everything) in realtime.

Bow if youre in the cloud, on a dataset, or joysticking so you dont care about up to date updates, so much larger possible.

Async extends that and basically doesnt queue up scans so you’re always live for so much large maps, but youll have to watch out after 200k that youre not going to fast and not getting a dense output, per standards for cartographer karto, etc. im not sure i know any synchronous slam implementations that are main stream for this lag reason but we spend it up really nicely to fit really any retail/not insane amazon warehouse size.