Replace FLANN w. nanoflann/Annoy and merge it with Features[2D]

[vpisarev]

Describe the feature and motivation

Approximate nearest-neighbor search is very important research and development area of today's AI technologies. Besides computer vision (e.g. in face or scene recognition applications), they are actively used for recommendation systems, e.g. https://www.benfrederickson.com/approximate-nearest-neighbours-for-recommender-systems/.

There are high-profile specialized packages for this task, mentioned in the above paper, such as Annoy, NMSLib, Faiss.

By comparison, FLANN, a fork of it integrated into OpenCV, looks very old and it's not maintained. It is suggested that in OpenCV 5, since we have an opportunity to clean the library, we will remove a separate module FLANN.

The following two options are suggested:

1. use single-header BSD-licensed nanoflann: https://github.com/jlblancoc/nanoflann/tree/master/include. It implements essential subset of FLANN functionality, so it should be a smooth replacement.
2. use Apache-2 licensed Spotify's Annoy library: https://github.com/spotify/annoy. Annoy contains just a few files and so it can also be put directly to OpenCV, without creating explicit opencv/3rdparty entry.

The upgraded 'mini FLANN' (nanoflann or Annoy) can be put to features2d where it logically belongs.

Probably, features2d should be renamed to features to reflect that it can be used not just for features extracted from 2D images. The current modern way to compute features is to run a deep learning model.

Additional context

No response

[crackwitz]

Suggestion: adjust title to something like

Replace FLANN implementation, move into features2d

At first glance, the current title looks like removal of the feature, not dissolution of a module and relocation of its APIs to another one.

Replacing the old code with something maintained, strikes me as technically sound. At this time, I have no further opinion on any of this.

[vpisarev]

@crackwitz, thanks, I changed the title