Fixed bug in batch example that did not add last set of measurements in file
Created incremental version that reads measurements as they come, associates with a smart factor (or creates new one), and optimizes.
Last example achieves same results on 200-pose example as batch example.
Currently loads kitti dataset after processing by VO code (stereo_factors.txt and camera_poses.txt)
Creates graph after N landmarks have been seen, runs it through SmartProjectionFactor
Verification of output is still needed; LM seems to optimize but poses don't seem to change.