Finalise Kalman filter

[clarenc3]

• [x] Fix bugs in path calc., geometry, etc.
• [x] Stabilise fitter
• [ ] Deal with unknown charge in a sane way
• [ ] Account for B field in propagation
• [x] Accurate error matrix calculation
• [ ] Get more info than momentum
• [ ] Do justice to Clarence' code

[LiamOS]

Am working on adapting this to the current 3D reco output on branch feature/3D-Kalman-fitter

[LiamOS]

PR #122 provides the basis of a stable Kalman filter. Currently lacks some of the niceties we want in the final version.