Moved a bunch of lookup logic into rlookupWild, which let me delete a
lot of special-purpose and flawed code. It is clearly heading toward
being properly-refactored, like the Racket implementation (and the
ESOP2016 paper's presentation) is. The performance problems may
interrupt this gradual evolution before it is complete, though:
hopefully I will be able to move to an explicitly memory-managed
scheme soon.