A small CPLD, e.g. XILINX XC9536 or the like, or something from Altera or Lattice might serve, and a bigger one could even help, and will take the place of all the latches, etc. so long as you get one with enough pins. Once you have 'em connected to the MCU pins, the rest is "inside," so it's transparent to the rest of the circuitry. I think you'll find that convenient. CPLD's can support bidirectional data paths, and/or, can easily control external data path logic if needed. A 100-pin TQFP, if you need that many pins, is about the size of your thumbnail and can do quite a bit. They're pretty cheap in TQFP, too.

You can build a sequencer that will steer/mux your data synchronously with your clock, and make everything fit the MCU timing.

RE