Add Gater classes

[sglvladi]

Summary

This PR adds a Gater base class which can be used to wrap hypothesisers, therefore allowing for the reduction (a.k.a gating) of the hypotheses before they are processed by data association.

Contributions

The main contributions/changes are outlined below:

• The concept of a Gater object has been introduced. Gaters wrap Hypothesisers in a similar way to how Feeders wrap Detectors. As such, the Gater class is defined as a sub-class of Hypothesiser. This idea has also been previously discussed here, albeit in a slightly different context.
• The FilteredDetectionsHypothesiser has now been renamed to FilteredDetectionsGater as it falls under the Gater class conceptually.
• A new DistanceGater has been introduced, which, uses a measure (e.g. Mahalanobis) to calculate the distance between a hypothesis' measurement prediction and the hypothesised measurement, and then removes any hypotheses whose calculated distance exceeds the specified gate threshold.
• As a consequence of the above, the internal gating performed by JPDA has been removed. The equivalent gating process can now be achieved by wrapping the PDAHypothesiser with a DistanceGater.

[sdhiscocks]

Looks good. We should definitely look to make that change for Hypothesisers and now Gaters to take multiple tracks, such we can wrap the classes as discussed in #145.

[jmbarr]

Neat. Like the fact that the 'principled' hypothesiser is now decoupled from the 'arbitrary' gating step.