Improve point trajectory estimation by aggregating across sources #349

[demoncoder-crypto]

This PR adds a new track fusion module that implements multiple methods for combining tracking data from different sources to produce more accurate trajectories. This addresses issue #271.

Key features:

• Multiple fusion methods: mean, median, weighted average, reliability-based, and Kalman filter
• Handles alignment of datasets with different time coordinates
• Provides a high-level interface through the fuse_tracks function
• Includes comprehensive unit tests
• Includes an example script demonstrating usage with sample datasets