Implement live synced view all the time

[kltm]

Depends, is related to, #14

[kltm]

Any work here needs to sync with the eventual default layout solution in #80.

[kltm]

Current sub-items will be dealt with here:

• for every model, there is a messaging "cubby" (hash) to store transient clobbering information
  • this should always be treated as an optional path needs to be in there
  • "layout" skimming can be passed as a function? do this internally to barista initially
• all of cubby is passed to a client on model messaging connect; client does what it will (other uses besides layout); after update, the current message stream should keep things in sync
  • not going to work, would likely end up broadcasting all models layout...maybe injection on model rebuilds? this actually makes more sense as I want to have that information as soon as possible and it would be likely to keep people in sync injection not possible on response: How to modify the response HTML data going back to the user? http-party/node-http-proxy#382
  • maybe bounce it through minerva and back with injection? that would still keep things in sync I'm now doubtful that this is satisfactory because barista will likely only ever have partial information, which would make request injection a bit silly in light of Implement method of storing a "default" layout of nodes (as moved/saved by editors) #80
  • this leaves us with a two-step system on the client: 1) render 2) request updated information for model from barista

Cubby information will inevitably accumulate until forming a black hole, taking barista with it. Eventually, some important cubby (e.g. layout) information will need to be stored (and initially retrieved from) the server model on initialization.

[kltm]

While layouts are syncing, merge and rebuild triggers are also needed.

[kltm]

Messaging system can now intercept and read outgoing models from the api and pass them out; the client can now catch those messages on callback and hook them back into the minerva manager, updating the models.