Implement BFMatcher and document AKAZE differences

[kalwalt]

This pull request introduces a new feature matching visualization example and implements a set of drawing utilities for feature detection and matching in the purecv library. It also implements BFMatcher and DMatch for descriptor matching. The main changes include a new example demonstrating ORB feature matching and a new draw.rs module for visualizing keypoints and matches, both with detailed documentation and LGPLv3+ licensing headers.

New Example and Visualization Utilities

Example and Usage

• Added a new example examples/match_features.rs that demonstrates ORB feature detection, extraction, matching (with mutual cross-check using BFMatcher), and visualization. The example loads examples/data/graf.png (a stitched image), splits it down the middle, detects and matches features, draws the matches, and saves the result to examples/data/out/graf_matches.png. Extensive documentation and usage instructions are included.
• Added examples/data/graf.png as test data for the example.

Feature Matching Infrastructure (src/features2d/matcher.rs)

• Implemented BFMatcher (Brute-Force Matcher) with support for NormType (including NormHamming for binary descriptors) and optional cross-check for mutual consistency filtering.
• Introduced the DMatch struct to represent individual descriptor matches.
• BFMatcher implements the DescriptorMatcher trait via the match_descriptors method.

Drawing Utilities for Feature Visualization (src/features2d/draw.rs)

Introduced a new draw.rs module implementing pixel-level drawing utilities:

• Functions for drawing lines (Bresenham's algorithm) and circles (midpoint algorithm).
• The draw_keypoints function draws circles and orientation lines for detected keypoints.
• The draw_matches function visualizes matches between two images by drawing lines between corresponding keypoints. Supports random or fixed line colors; passing None for the unmatched-keypoints parameter suppresses their rendering.
• All functions are documented, and the module includes an LGPLv3+ license header with classpath exception.

Module and Library Updates

• Updated src/features2d/mod.rs to expose the new draw and matcher submodules and re-export their public API.
• Updated src/lib.rs accordingly.
• Added tests in src/features2d/tests.rs covering the new functionality.

---

Note: This example differs from the OpenCV AKAZE matching and tracking tutorials in that purecv implements ORB and FAST rather than AKAZE, and matches are filtered using mutual consistency (cross_check = true) rather than homography estimation.