Explain queries may work better with larger datasets

[fabioperrella]

Could it be an idea to try to install a "representative dataset" for testing the gem? For example using a fuller list of Brazilian municipalities + a longer list of people names, say generated with Faker. I suspect it is really dependent on the table size, both for MySQL and PostgerSQL, what the query planner will plan.

Indeed, I also think that the results would be better if using more data, but this will not be the case when it analyses queries running on unit tests, for example.

I will create an issue with that, maybe an idea can be born influenced by that!

Originally posted by @fabioperrella in #13 (comment)