Swap gamma for Drude and Lorentzian terms in epsilon fitting parameters for Au from Johnson and Christy and update Titanium

[oskooi]

The material fitting parameters for Au at visible wavelengths from Johnson and Christy provided in D. Barciesi and T. Grosges, J. Nanophotonics, Vol. 8, 083097 (2014) contain an error described in the errata in which $\gamma_D$ and $\gamma_L$ (the loss rate for the Drude and Lorentzian terms, respectively) are swapped. Unfortunately, the current values for these fitting parameters in the materials library do not contain the correction described in the errata. This PR updates these fitting parameters with the correct values.

As validation, a plot of the real and imaginary parts of $\epsilon$ using the updated parameters matches the plot in Figure 1(a) of the original paper. (The abstract in the errata states "Fortunately the curves in Ref. 1 [original paper] are correct.") The script used to generate this plot is provided below as a reference.

Also, because of a likely error for the fitting parameters of Ti in the errata, this PR uses the fitting parameters from the original paper with the corrected $\gamma$ terms.

cc @kombatEldridge

"""Plot the complex epsilon for Au in the materials library."""

import matplotlib
matplotlib.use('agg')
import matplotlib.pyplot as plt
from meep.materials import Au_JC_visible
import numpy as np


wavelength_um = np.linspace(0.4, 0.8, 50)
frequency = 1 / wavelength_um

epsilon_r = [Au_JC_visible.epsilon(f)[0][0].real for f in frequency]
epsilon_i = [Au_JC_visible.epsilon(f)[0][0].imag for f in frequency]

um_to_nm_scale = 1000

fig, ax = plt.subplots()
ax.plot(
    wavelength_um * um_to_nm_scale, epsilon_r, 'bo-', label='$\Re{(\epsilon)}$'
)
ax.plot(
    wavelength_um * um_to_nm_scale, epsilon_i, 'ro-', label='$\Im{(\epsilon)}$'
)
ax.set_xlabel('wavelength (nm)')
ax.set_ylabel('$\epsilon$')
ax.set_title('Au_JC_visible from materials.py')
ax.legend()
fig.savefig('epsilon_Au_JC_visible.png', dpi=150, bbox_inches='tight')