Support big files

Loading big files (e.g. 100MB+) is actually breaks the application. Optimize the geppetto model building in the backend.
The logic to optimize is in the `NWBModelInterpreter` class and delegated mappers.
Test with 1GB+ real files, real files here: https://neurodatawithoutborders.github.io/exampledata

Start with the backend NWBModelInterpreter class and its delegated mappers to trace model construction for large NWB files. Test with the linked real example files, including 1GB+ files, and verify that the application can load them without breaking while preserving the expected model output.

Written by the indexing model from the issue text.