The file size of 16-bit scans is not really an issue on modern computers. The bottom line is, if you're planning to do significant colour or tonal correction on your scans and you care at all about quality, you have to have 16-bits to play with. I know that sounds absurdly prescriptive, but with experience it quickly becomes self-evident - it's not voodoo, it's simple mathematics. By all means store the finished scan in 8 bits though (that is, if you're confident that you can never improve it at any time in the future
).