Add auxiliary particle filter

[hidmic]

Feature description

Connected to #424. Perhaps #187 too. Auxiliary particle filters are relatively simple variation over bootstrap particle filters (BPF) in that they steer the BPF towards regions of higher probability by pre-weighting and sampling particle indices based on the latest measurements, then focusing on the particles that matter for predict-update cycle. There is no resampling step. See here for an amenable explanation.

It should be easy to add support for generic APFs in Beluga.

Implementation considerations

Beluga's ranges based design comes in handy here. We can implement APFs by changing the order of particle cloud actions and adding some index sampling view.