Move further towards KdTreeNanoflann Part B

[mvieth]

With this PR, it is now possible to build most of PCL without having FLANN installed (apps, examples, test, benchmarks, and some tutorials still need FLANN and the kdtree module that depends on FLANN). It is still highly recommended to have either FLANN or nanoflann installed. If both are available, PCL auto-selects KdTreeNanoflann by default (faster than KdTree). I have tested with only FLANN installed, with only nanoflann installed, and with neither FLANN nor nanoflann installed (all scenarios build successfully). The CIs additionally check the scenario where both FLANN and nanoflann are installed.

This PR

• adds a new overload for pcl::search::autoSelectMethod where the PointRepresentation to be used can be specified (the other overloads use the default point representation). This overload uses KdTreeNanoflann, KdTree, or BruteForce, depending on what is available.
• changes KdTreeNanoflann to not inherit from KdTree but from Search directly (KdTree is not available if FLANN is not found)
• further moves from pcl::search::KdTree to autoSelectMethod in the registration and recognition modules
• removes the strict requirement of most PCL modules on the kdtree module (no longer required)
• adds setPointRepresentation in pcl::search::Search (overridden in several subclasses, e.g. KdTree and KdTreeNanoflann), and makes it return whether it was successful and the point representation will be used, or not.
• The if-else-statements in registration.hpp and correspondence_estimation.hpp are more complex than I would have liked, but this is the only way to cover all possibilites (e.g. point_representation_ given by user or not, tree_ given by user or not, indices given by user or not, ...)
• A comment on ppf.hpp/cppf.hpp/rgbppf.hpp: These classes do not use tree_, and I think it is technically not necessary to fill tree_ with BruteForce there (would otherwise be filled in Feature::initCompute ()). It only has the minor advantage that BruteForce does not perform any work preparing for searching (like setting up a kdtree, which would be wasted time, since the tree is unused)

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[Copilot]

Pull request overview

Copilot reviewed 41 out of 41 changed files in this pull request and generated 5 comments.