The 2116 has limited memory for leaning. It has seperate memory for Key Moves. If you can define the functions you want as Key Moves, you probably can handle more functions as Key Moves than you could as learned signals. Since the memory is seperate, you definitely could fit more as a combination of Key Moves and learned signalsthan with either alone.

The RC5 protocol has a "toggle bit". A learned signal can't get the toggle bit right. A key move does. On most devices, if you get the toggle bit wrong, the device won't recognise the use of the same key twice in a row. (Before any key can be used again you'd need to use some other key. That can get real annoying if you try to enter the number 11, assuming a device where entering two digit numbers means something).

I don't know the OmniRemote format at all and can't guess it from that fragment. I do know RC5 well enough that I think I'd recognise it in some unknown encoding if I saw a few complete samples. But I thought you said you found Pronto info on this brand. That would be an easier way to find out what the signals are.