the pentium2 used this design because it did not include L2 cache in the actual cpu die as all later cpu's did.
the pentium p54c and p55c had their L2 cache on the motherboard, at system bus speed. the pentium pro had the L2 cache in the ceramic cpu package, at full processor speed (233/266/300mhz), but this was expensive to produce.
when they brought the p6 design to mainstream production, cheaper, half cpu speed L2 cache chips were used along with the cpu core in its own package, with both on a daughterboard in a common heatsink assembly called a slot-1 cpu. this was both cheaper to produce than the pentium pro and faster than the L2 at bus speed design of previous cpu's.
when die shrinks and memory technology advanced sufficiently to allow L2 cache to be integrated into a single cpu die along with the traditional cpu (coppermine pentium 3) it became cheaper to manufacture just the one chip in a pinned package again, as no external cache chips needed to be included with it.
this had nothing to do with a technology limitation, and everything to do with cost. electrically, slot-1 and socket-370 were virtually the same, with similar pinouts.
