Roofer v1.x overview

• Develop API
  • Investigate more comprehensive API for individual steps from reconstruction pipeline (?)
• Refactor code. Focus on common datatypes, minimise code duplication etc.
  • #11
  • review attribute handling
• Improvements to handle very large input data without needing a separate pre-tiling step
  • stream reconstruct output as CityJSON Sequences to disk or to another program
• Improve error handling roofer app
  • keep track of which batch/tiles have completed during processing in a 'project file'. So that we can pickup immediately from a crashed tile if we restart (add a cli option to enable/disbale this behaviour). This is to handle crashes that do not throw exception. To identify a run use a hash of all input options/parameters.
  • option to output intermediate data from crop phase for subsequent debugging/reconstruct
• misc features
  • simplify parameters. Remove unnecesary ones, automatic tuning (eg based on point density) were possible.
• Automated testing
  • Unit tests

No files, tests, or entry points are named; begin by breaking the roadmap into a single scoped item and locating the reconstruction pipeline or roofer CLI entry point. The listed goals range from API and refactoring to large-input streaming, restart state, error handling, parameter simplification, and unit tests, so completion is not defined for this issue as written.