If you had a 300a@450 or a 366@550, then a 1gig celeron2 would be the obvious answer, but you really wouldn't notice much difference over what you have now..
Once you have a cpu in the 800mhz and up range, it's difficult to justify further upgrading on a bx board. Problem is that the p3's are overpriced and the celly2's only go to 1100. They will overclock somewhat, depending, check some of the dedicated O/C site databases to evaluate your chances. There's some info that tualatin celerons will run on modified slotkets, check the same sites for more info.
While the powerleap stuff works, it's expensive, you'd get better performance spending the same money (or a little bit more) on a new board, memory and cpu.