Steps needed to prepare measurement engine to be extracted from PartSeg

1. Update migration engines to allow declare migration next to data structure definitions (almost done) feat: Better serialization mechanism allow for declaration data structure migration locally #462 https://hackmd.io/@GytvakMcS_WIFerhsG8H3w/H1dhn-wpF.
2. decide where migration engine should be available (maybe nme aka napari-migration-engine`)
3. Allow to register function-based measurement (currently accept only class-based and comes from python 2.7 where there is no annotation). The best will be if it will be possible to bedone with npe2. Also may need:
   • Update hash function for better-caching results during calculation
   • Better area definition that will cover not only Labels but also Shapes etc.
   • maybe Better Per component definition. Currently, the measurement could be calculated per Label component, globally, or Mean from Per component. Adding more option may reduce needs of postprocessing but complicate the interface.

For performance reasons measurement engine uses cythonized calc_bounds function
https://github.com/4DNucleome/PartSegCore-compiled-backend/blob/494dee185a3a965f90ba06aacdd1a896d0681fe1/src/PartSegCore_compiled_backend/utils.pyx#L13

This allows a clipping array that gains significant performance speedup but also introduces a binary dependency.

I will update this issue.

@kevinyamauchi