* split crate into library and binary
criterion can not benchmark binary crates
* Add store benchmarks
* Add more, not reproducible benchmarks
The benchmarks are currently hardcoded to use a specific system
on my computer. To run the tests yourself, change QUERY_DERIV__OLD in
benches/common.rs
In the future, I would like to pull a specific copy of the
nixos sqlite db from e.g. a github repo to make the benchmarks
reproducible
* Do not hardcode system derivations to use in benchmars
By default, the oldest system is compared to the current system
A specific system can be specified using environment variables (see
benches/common.rs)
