Added a skeleton benchmark suite + mockup queries

[tuomohopia]

Based on the brief discussion here I built up a skeleton for microbenchmarking query rendering. As it is, it's probably not useful but meant to act as a foundation to build a real benchmark suite on.

Run via stack bench.

Queries with parameters

What I don't understand is how the query actually gets composed with its params when running manipulateParams_ ....

There's lots of extra noise here, mostly because I wrote some facilities to auto-generate manipulations with different params based on the data types.

The idea was to facilitate benchmarking queries rendered with auto-generated parameters. I mean benchmarking building up the full ByteString that actually gets sent to Postgres.

I don't expect this to be merged, I just want to know if I'm heading down the right path trying to help with the slow query rendering issue or not.

[echatav]

Fantastic! Thanks for contributing this. Just noting some ideas to add later:

benchmark time to connect
benchmark running manipulations and queries
benchmark generic decoding and encoding of rows, params, composites and enums
benchmark hand-written decoding and encoding of rows, params, composites and enums
benchmark basic value decoder and param encoders
benchmark very large / complex query rendering
experiment with using Builders instead of ByteStrings
experiment with applying strictness and forcing