Fix conversion of matplotlib contour lines

[robertoffmoura]

mpl_to_plotly doesn't render contour lines correctly. Matplotlib packs each contour level into a single path containing several disjoint subpaths, so the converted figure connects them with segments that shouldn't exist, and closed rings are left open because the CLOSEPOLY vertex is dropped on export:

• separate branches of the same level get joined by diagonals
• every closed loop misses its final segment

Fix: path collections with no face colors (line collections, which is what ax.contour produces) are now drawn as line traces instead of filled polygons:

• disjoint subpaths inside a path are separated with None, so plotly does not connect them
• subpaths ending in a Z code have their first vertex appended to close the ring
• multi-vertex path codes (C, S) consume the correct number of vertices while parsing
• degenerate paths (fewer than 2 vertices, e.g. the empty path some contour sets carry) are skipped
• x-values are converted to date strings on date axes, and per-path edge colors/linewidths are applied in order

Snippet to reproduce:

import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
import numpy as np
import plotly.tools as tls

x = np.linspace(-2, 2, 120)
X, Y = np.meshgrid(x, x)
Z = np.sin(3 * np.sqrt(X**2 + Y**2))

fig, ax = plt.subplots()
ax.contour(X, Y, Z, levels=[-0.75, -0.25, 0.25, 0.75], linewidths=2)
fig.savefig("contour_mpl.png")

p = tls.mpl_to_plotly(fig)
p.write_image("contour_plotly.png")

matplotlib	plotly before	plotly after