Fix #293: State to device side effects

[sacredvoid]

Summary

Fix: Resolve SimState side effects in device/dtype conversion
Problem: SimState.to(), concatenate_states(), and initialize_state() were modifying input states instead of creating new instances.

Solution:

Fixed _state_to_device() to use copy.copy(vars(state)) instead of direct reference
Updated functions to create new instances without side effects
Key Change:
Impact: Fixes Issue #293, eliminates unexpected state mutations, maintains backward compatibility.
Testing: Added test_fix_demo.py to verify all operations preserve input states unchanged.

Checklist

Before a pull request can be merged, the following items must be checked:

• Doc strings have been added in the Google docstring format.
• Run ruff on your code.
• Tests have been added for any new functionality or bug fixes.

[sacredvoid]

Hi @orionarcher and @thomasloux, saw your message on the slack channel and wanted to make my contribution, let me know if any edits are required!

[thomasloux]

by the way, not sure if we need this kind of tests in the repo

[orionarcher]

Thanks @sacredvoid! Welcome to the project. One small request but otherwise looks good!

[orionarcher]

Let's not copy here. The behavior of .to() below is such that it will create a copy if the device is different.

[orionarcher]

I'd agree with @thomasloux that we can get rid of most of the tests. I'd just test .to() and not the downstream behavior.

[sacredvoid]

@orionarcher hey, i've updated the PR with requested changes, let me know if it's good to go, thank you

[orionarcher]

Nice job @sacredvoid, LGTM! Will merge when tests pass.

[thomasloux]

Hey, I would remove the tests and modify the description. Now, the IS to change the input state. I would also directly return state to prevent create a new dataclass object with all its variables pointing to another dataclass object