More careful management of hash table allocations

[wesm]

per https://github.com/pydata/pandas/issues/4491

we may consider a fixed-size memory pool (which could be managed with an LRU stack) for hash table data to avoid excess internal index hash tables

[chris-b1]

xref https://github.com/pydata/pandas/issues/14273 from @ssanderson - this particular case could still be improved in pandas 1.0, but a good example of where hash table size can be problematic.