Added new function to hash any PostgreSQL data type

[ozgune]

Added new function called hll_hash_any that can hash any PostgreSQL data type. Also added corresponding regression tests for hll_hash_any.

[ozgune]

I added a new hashing function that resolves the PostgreSQL data type dynamically, and dispatches to the corresponding hash function.

I also measured the performance overhead of using this function over columns in the TPC-H data warehousing benchmark's lineitem table (scale = 1GB, 6M data points). I'm including the performance numbers here for documentation purposes. As a performance baseline, TPC-H query #1 completes in 8s on my machine.

Data type | Direct hash call | Any hash call
Text 3.3s 4.6s
Text (low cardinality) 6.4s 7.6s
Bigint 2.1s 3.4s
Integer 2.1s 3.4s

hll_hash_any therefore has a performance impact. Still, it can be used when the user doesn't know the underlying type in advance, or more importantly when the column data type isn't supported by one of the hash functions.

[ghost]

@ozgune I'll take a look at this today. Thanks for the PR!

[ghost]

Thanks, @ozgune!