I guess that your animation sequence is likely to evolve rapidly and continually but your basic mechanism for reading and executing instructions is not. On that basis I would be inclined to go for a text file on the SD card that can be shoved into a PC and reprogrammed quickly and easily in the field without having to build and upload new software. (This also gives you an obvious way to archive your animation sequences and switch between them easily etc.)
For the text file format, the simplest format would be ASCII CSV with each record containing the sort of info you listed.
It seems unlikely that the number of cues or rate of executing cues is going to be big enough to be an issue, but if it is then figuring out how to generate those cues is going to be a far more important problem.