Add par_maplist/{2,3,...}

[mndrix]

Implement a parallel version of the maplist predicates which uses a pool of workers (one per CPU) to evaluate each goal. The architecture is roughly:

• create a message queue from which workers read pending jobs
• create a message queue to which workers write finished jobs
• create one worker thread per CPU
• enqueue Index-Element values to pending job queue
• dequeue Index-Result values from finished job queue
• sort resulting pairs by Index
• return list of Result values

I have a simulation job whose core operation is performing a maplist/3 to calculate many independent values. It would be nice to use all cores for this operation.

[mndrix]

library(thread) already has concurrent_maplist/{2,3,4} so this is not needed.