Added support for EMESimulation in ModeSolver plugin

[caseyflex]

Previously, we could add a ModeSolverMonitor to an EMESimulation, or we could convert the scene to a Simulation to use the ModeSolver plugin. Now, we can directly pass an EMESimulation to the ModeSolver plugin.

The change was suggested by @tomflexcompute here: #1633

It will also facilitate the planned ModeSimulation classes.

[tomflexcompute]

That's a great addition. Thanks @caseyflex . I wonder if one uses the local mode solver to solve for the modes and dose smatrix_in_basis for example with the mode solving result, would the accuracy be noticeably worse?

[tylerflex]

make sure to update the changelog to put this in the Added section of Unreleased instead of 2.7.1

[caseyflex]

That's a great addition. Thanks @caseyflex . I wonder if one uses the local mode solver to solve for the modes and dose smatrix_in_basis for example with the mode solving result, would the accuracy be noticeably worse?

I actually imagine it wouldn't matter so much. The mode profile should be similar, only the details and the n_eff would change a bit. And n_eff is very important for propagating EME modes but perhaps not so much for port modes.

[momchil-flex]

Looks good. Does it really facilitate the ModeSimulation refactor?

[caseyflex]

Looks good. Does it really facilitate the ModeSimulation refactor?

We would like to pass a ModeSimulation to the ModeSolver as well, without trying to come up with a way to convert it to an FDTD simulation.

[caseyflex]

Looks good. Does it really facilitate the ModeSimulation refactor?

We would like to pass a ModeSimulation to the ModeSolver as well, without trying to come up with a way to convert it to an FDTD simulation.

(even if this isn't the primary web api, it will help internally to support it, and it may be helpful for users as well)