Add plot1D function for visualizing layout of 1D simulation

[oskooi]

Adds a plot1D function for visualizing the layout of 1D simulation. This mirrors the existing plot2D.

PEC detection. A wall is PEC when the cell boundary is metallic — i.e. no k_point was given (Meep's default) or set_boundary(..., mp.Metallic) was called. Since fields.boundaries is only exposed to Python as an opaque SWIG pointer, Simulation.set_boundary now records what it set in self._boundary_conditions (cleared by reset_meep), which plot1D consults. This makes the tutorial setup in antenna_pec_ground_plane_1D.py (Bloch k_point + explicit metallic walls) plot correctly.

Everything is customizable through the same *_parameters dict / show_* flag convention as plot2D (index_parameters, pml_parameters, pec_parameters, source_parameters, monitor_parameters, legend_parameters), and the dispersive-permittivity frequency defaults to the first source's frequency (that lookup is now the shared get_default_frequency helper, also used by plot_eps).

An example of the output (involving a dielectric slab above a PEC ground plane) produced by plot1D is shown below.

Files changed

• python/visualization.py — plot1D and its plot_1d_* helpers, 1D default-parameter dicts
• python/simulation.py — Simulation.plot1D with full docstring; boundary-condition tracking
• python/meep.i — export plot1D from the package
• python/tests/test_visualization.py — test_plot1D covering legend contents, artist counts, PEC vs. Bloch-periodic, Absorber, set_boundary override, customization, and the non-1D ValueError
• doc/docs/Python_User_Interface.md{, .in} — API docs (inserted exactly the block generate_py_api.py emits)

Verification

• python/tests/test_visualization.py: 3 tests pass, serially and under mpirun -np 2
• Checked against variants: PML both/one side, Absorber, Bloch-periodic, explicit metallic wall, extended sources, dispersive material (Si reads 3.478 at 1.55 μm), nonzero geometry_center, dimensions=1 and unset-dimensions 1D cells

[stevengj]

Rather than add a Python mirror of the boundary information, I'd rather just add a method to fields:: that exposes this information, e.g. a fields::get_boundary function (mirroring fields::set_boundary) or something like that.

[oskooi]

Rather than add a Python mirror of the boundary information, I'd rather just add a method to fields:: that exposes this information, e.g. a fields::get_boundary function (mirroring fields::set_boundary) or something like that.

The Python Simulation._boundary_condition object is removed. The boundary condition is now read from C++ using a newly added fields::get_boundary.

Two things worth flagging:

1. Fixed a deadlock. get_boundary can call init_sim(), which is collective. The original plot_1d_pec had an early if not mp.am_master(): return ax before the query, so under mpirun -np 2 only rank 0 entered init_sim and the run hung. The query now happens on all ranks before the master-only drawing, with a comment saying why.
2. The plot got more accurate. Querying C++ revealed that the original Python mirror was wrong for the case of set_boundary(mp.Low, mp.Z, mp.Metallic) on a Bloch-periodic cell which makes both walls metallic, because fields::set_boundary forces the opposite side to match when it was Periodic. The old code drew PEC only on the low side. The updated schematic below confirms this behavior.