avoid wasted space following the header in small chunks

[thestinger]

A 4MiB chunk has room for 64 of the 64kiB slabs, but the first 64k is currently entirely lost due to the 8 byte chunk header. An alternative is for each slab to have 8 bytes of padding at the start so that a slab would fit neatly after the chunk header. It would be way too expensive to pay for a branch to handle this special case, but losing some extra bytes per slab isn't a big deal.