Question regarding spherical coordinates and a hard code focal length

[Tram-Tran]

Hi, thank you so much for providing such a great work. I have a few questions related to the distance computation process.

1. Why the distances are computed in spherical coordinates and then converted back into Cartesian instead of computed directly in Cartesian coordinates?
2. The focal length is the most important value to convert camera-to-world coordinates in a pinhole camera model. I observed that it is currently set as a fixed number. Is there any specific reason? Do I have to change it to adapt different datasets? (it's not in the configurable variables anyway)

Thank you for your time and consideration.