There is only one thing in the 010 that is faster than the 000: it has a six byte (!) loop cache which means that the smallest possible loop on a 68k, a short instruction followed by a dbcc instruction, will be faster than on a 000. Everything else is identical speedwise. That one instruction became a privileged instruction is hardly relevant, it has always been a privileged instruction in all 68k since the 010 and that was never much of a problem.
The mini-loop is very rare in normal code, it was intended for speeding up hardware polling and perhaps a mem-copy. In my opinion spending 5$ for an 010 is 5$ wasted.
