While I do not recall the EXACT calculation for "load average", my recollection is that it is the average number of processes waiting for either CPU or IO (usually disk). So it is very common to see a high load average with low CPU utilization.

The RAID statistics in /proc/ are there because the RAID code is deliberately self-throttling while rebuilding. The statistics serve as a tuning indication only, in case one wants to slow it down even further.

Nearly all block I/O drivers (including IDE) keep similar numbers in the kernel, and there may exist userland applications for displaying them. The IDE driver doesn't have the code bloat that would be needed to calculate & display them in /proc/ (but these days it does have lots of other bloat..).

Cheers