Add support for serializing to JSON

[MicahGale]

Due to the time it takes to read a model into MontePy, it is helpful for users to be able to serialize the model and write it to file to make loading in the future easier. Pickling is possible, but does come with a huge security vulnerability when sharing files. Instead a way to support serialization that does not allow code execution is needed. So JSON should be supported.

Some thoughts:

1. Most objects are not immediately writeable to JSON.
2. The "correct" way to implement this is by implementing a default function that is passed to the load/dump function. This doesn't seem the cleanest as it would be nice for JSON to call a dunder function that an object implements so json.dump can just be called on an object.
3. The defacto standard is to implement a to_json method, and sometimes a from_json.
4. The syntax tree for each object is it's most important information to keep.
5. There are two ways to save to JSON.
   1. An internal representation of the state for the MontePy object e.g., {"cell": {"number": {"value": 120, "token": "0120"} ...}
   2. A JSON version of an MCNP input file. The goal with this would be to push MCNP in this direction: e.g., {"cell": {"number": 120, "mass_density": 10.0, importance: {"n": 1.0...}}}

How to implement:

1. Implement MCNP_Problem.export_to_json() for creating a JSON formatted Input File
2. Implement MCNP_Problem.to_json() for serializing the object state to json.
3. Maybe we will want to create a generic MCNP_Problem.serialize that will just generate a dictionary that can then be saved however the user wants.
4. Need to implement montepy.from_json.

[MicahGale]

In GitLab by @tjlaboss on Dec 14, 2023, 20:22

I like serialize() as a public method so that MontePy users can select the language that serves their needs: JSON, YAML, XML, etc. Adding methods like .to_json(), as Pandas does, are also nice for UX.

[MicahGale]

to_json would be for a complete serialized dump? For MCNP input json we could do to_mcnp_json maybe?