Apple already made a strong commitment to AirPlay, so they already had a focus on building a fast, smooth, low power encoder that could encode the entire screen. Once encoded, the stream is probably only a few Mbit/s, a data rate that can easily be transmitted with single ended protocols like SPI. Almost all SoCs already have multiple SPI buses so no need to change any hardware.
Their solution may seem inelegant to some, but I think it is great. They managed to support a feature with almost zero hardware costs on the core device (depending on how you assign the lightning connector). A feature that I'm willing to bet only a very small % of all users will ever use. That is a big win, not having to dump extra hardware into a device that only 5% of all users will ever activate does great things for your margin and design flexibility.
MicroUSB+MHL accomplishes the same thing quite well on the current crop of Android phones. Why Apple chose the method they did is still quite odd.