make bug search page when possible + add yield support

searching large data sets takes ages. to reduce load spikes and make the
iteration easier to follow we'll page requests (i.e. break them up into
multiple request sets). each page gets iterated on its own and will yield
the Bug instances if a block was given.
put together this will allow us to iterate bugs more interactively, spread
out load and also reduce the memory footprint by keeping less unnecessary
instances in memory