Imu Interface initialization and closing

[rikba]

From a user perspective I find it difficult to know which functions I have to call when loading an IMU, e.g., init after object creation, init and burst CRC activation in factory, self-test in node init. Where does the IMU call its close() function?

I think it is worth to sit down and make a clearer architecture on how a "sensor" is added and removed. I can also see that any routine that cannot fail (or we have no method to react to a fail) can go to the IMU object constructor or destructor and hiding the functionality from the public interface.

[4c3y]

I think we could change the ImuInterface::init() signature to
ImuInterface::init(const std::unordered_map<std::string, std::string> &config) to allow initialization of imu-specific settings, which could then also be loaded from a yaml file. With this change we could activate crc and burst and also execute the self-test from the ImuInterface::init function. We could then also remove the ImuNode::init() function.

The close function was never called and was part of bug #6. In #14 the ImuInterface::close() function is called after ImuNode::run_node is set to false by the SIGINT handler.