document special_kz algorithm

[stevengj]

Would be good to have a document that writes down exactly what equations we are evolving in this case.

[stevengj]

The basic algorithm is documented here:

meep/src/step_db.cpp

Lines 121 to 133 in 4e1802b

	/* In 2d with beta != 0, add beta terms. This is a trick to model
	an exp(i beta z) z-dependence but without requiring a "3d"
	calculation and without requiring complex fields. Looking at the
	z=0 2d cross-section, the exp(i beta z) term adds an i \beta
	\hat{z} \times cross-product to the curls, which couples the TE
	and TM polarizations. However, to avoid complex fields, in the
	case of real fields we implicitly store i*(TM fields) rather than
	the TM fields, in which case the i's cancel in the update
	equations. (Mathematically, this is equivalent to looking at the
	superposition of the fields at beta and the timereversed fields
	at -beta.) The nice thing about this is that most calculations
	of flux, energy, etcetera, are insensitive to this implicit "i"
	factor. For complex fields, we implement i*beta directly. */