Feature Request: Optionally Store MST and Condensed MST in HDbscan Struct

[alteredoxide]

For enhanced functionality and analysis, it would be beneficial to have the option to store the Minimum Spanning Tree (MST) and its condensed version within the HDbscan struct, along with a few new supporting methods. This would be similar to features offered by the python hdbscan library.

Some direct and indirect benefits of storing the MST:

• Explore clusterings with different configurations using minimal computation: min_cluster_size is only used when generating the condensed tree.
• Analyze and explore the MST to:
  • visualize data
  • automate parameter tuning
  • analyze edge weights and connectivity
  • compare other MSTs
  • leverage the MST for other algorithms
• Explore the hierarchy of clusters:
  • this one is perhaps more applicable to the condensed tree.

I've showcased an optional MST storage in a fork; please review the changes here. Let me know if this feature is of interest to you. If so, I can also add in optional storage of the condensed tree and create a PR.

I believe this enhancement can provide more versatility to the library. I hope to collaborate and refine the feature based on feedback.

Thanks for your work on creating this great library, btw!

[hadronzoo]

This would also be useful for implementing FISHDBC (Python implementation here).