Implement thruster control

[whymarrh]

Subtasks:

• Write thruster values over I²C

[whymarrh]

@cal-pratt can we break this up into a bunch of small pull requests to avoid unrelated changes

[whymarrh]

I've removed Joystick input and serial from this issue, as this can be thrusters via I²C only

[cal-pratt]

The branch 2016/thruster-communications contains the logic required for controlling the thrusters.

I wasn't quite sure how to use the Schedulers so I've been sticking to just the rx observables, but I think this branch is a lot more sane than the system-manager branch.

Within the entrypoint class Rov there is a static class HeartbeatObserver which listens for HeartbeatValue types from the EventPublisher. The HeartbeatObserver uses the time passed since the last HeartbeatValue and the operation status of the value to determine if the rov should be active. This is then chained with Observable.interval

final Observable<Boolean> operationObservable = 
    Observable.interval(SLEEP_DURATION, TimeUnit.MILLISECONDS)
              .map(l -> heartbeatObserver.isRovOperational());

This Observable<Boolean> is subscribed to by objects which implement Observer<Boolean> in the control package (objects which determine the next-state of the rov hardware) and io package (objects which communicate directly rov hardware). If the boolean they are passed in the onNext method is false then these objects will attempt to set the rov to a safe state, otherwise they will function like normal.

[whymarrh]

Sounds like progress! Is that branch ready to become a PR?

[cal-pratt]

I think there might be a couple minor issues with the output in ThrusterConfig6. The algorithm was based off of what Calvin gave me on Saturday, so we probably need to make some test cases to verify the results.

Apart from that, the functions used in I2CThrusterGroup have only been tested on one thruster. So they'll need to be manually tested on an i2c line with multiple thrusters connected.

[whymarrh]

Okay. If you're comfortable with that branch, there might be merit in opening a PR and using that to track the progress of the remaining tasks (e.g. tests and whatnot).

[cal-pratt]

The current state.