GMSO needs some optimization

[uppittu11]

My water example is extra slow in the atomtyping step (cell 8). Some optimization is necessary to scale up to a reasonably sized system (10000 water molecules).

https://gist.github.com/uppittu11/c421280913edee7e3a56ce09a7d3c2d3

Related to #144

[umesh-timalsina]

use update_types flags with sage whenever necessary. You will be able to deal with latency.

[rmatsum836]

Reopened due to some further discussion we had today. The update_ methods are O(n^2) because of the nested for loops. A couple thoughts we had:

• update_topology should only be called once during the initialization and atom-typing step
• In add_connections we already check for sites in connection.connection_members but aren't in topology.sites
• Though they should only get called once, the update_ methods are still time consuming for larger systems.
• Consider renaming update_ to something like check_, since in this method we are really performing a check to make sure all of our sites and connections have been added.