Benchmark Beluga NDT-MCL against a groundtruth signal

[hidmic]

Feature description

Connected to #93. Depends on #312, #313 and #348. In addition to tests and micro benchmarks, NDT-MCL nodes' performance must be compared against a groundtruth signal to prove it's correctness and precision.

Implementation considerations

Pick a pair of suitable datasets and benchmark 2D and 3D using LAMBKIN.

[serraramiro1]

@hidmic @nahueespinosa @glpuga
Based on our conversations, and us wanting to diverge a bit from nav2_amcl and the difficulties of making "fair" comparisons between different sensor models and map representations, I've changed this ticket to track the benchmarking of the model against a groundtruth signal instead. LMK what you think.