Unify `ModeSolver` with abstract simulation architecture

[tylerflex]

Feel like our mode solver might need a bit of a makeover. Perhaps we could try a new plugin to replace the existing one? but still support existing for backwards compatibility?

[daquinteroflex]

One thing I'd like to propose is that we need to write the equations in latex before implementing them in python so that the user can visualise what's happening underneath with clarity of solver limitations.

In particular I think we should properly document these methods: https://github.com/flexcompute/tidy3d/blob/pre/2.6/tidy3d/plugins/mode/solver.py

Happy to sit down and translate them to latex if we want to keep the existing implementation and just document it better for users. Or even write a little plugin explanation docs.

[tylerflex]

Yea a user had a similar comment saying that they wished we had more information about the equations we were solving in the solvers. It would be great if we could add that info to the solver docstring or to the run() method

[tylerflex]

I'll close this to just have a single issue #1523

[daquinteroflex]

Whoops sorry forgot this one existed, would have built onto this

[tylerflex]

It's fine, I think what we can do is if the 3.0 mode solver is good, we can just support it natively as a new plugin.

[momchil-flex]

Just a note that I think we'd move it out of plugins and into the core components on par with all other solvers.

[tylerflex]

yea I agree. although I think there's a possible design where we keep it as a plugin but just dont have it accept Simulation.