New Design for `config` system

[StFroese]

Currently the config is one python file config.py making it a Singleton-like module object:

from MCEq import config  # same object everywhere

This definitely comes with the advantages:

• You can directly create python objects in the config
• You load the config once and reuse it everywhere

But it also comes with disadvantages:

• The config is an implicit global space, changing it somewhere changes it everywhere
• Hard to override in tests because it changes the config for all instances
• Multiple instances of MCEqRun share the same config. Can’t run two configs side-by-side
• No validation of the correct input types. User can do anything to break it

Proposed solution:
Change to a pydantic config system.

from pydantic import BaseModel

class MCEqConfig(BaseModel):
    kernel_config: str
    e_min: float
    e_max: float
    standard_particles: list[int]

import yaml
from config import MCEqConfig

with open("config.yaml") as f:
    data = yaml.safe_load(f)

cfg = MCEqConfig(**data)

mceq_lowE = MCEqRun(config=cfg_lowE)
mceq_single_particle= MCEqRun(config=cfg_single_particle)