How to approach supporting alternative drivers

In the README (and I think I remember on Twitter) you mentioned being open to PRs for alternative SQL implementations. At $dayjob we use MS SQL Server and so I thought I might try my hand at a PR to add support for that. Before getting started, I thought it would be useful to see if you had any thoughts / opinions on how this should be done.

## Some current code to think about

There are at least a few things worth thinking about. The first is how we would specify the driver name, e.g. currently `"pgx"` here: https://github.com/felixge/sqlbench/blob/a7422d928ac6fd10ee4e91e52619da3be738ca0e/main.go#L77

And how it relates to the connection string here: https://github.com/felixge/sqlbench/blob/a7422d928ac6fd10ee4e91e52619da3be738ca0e/main.go#L35

There's also the EXPLAIN .. FORMAT JSON duration thing which I assume has equivalents in SQL Server and others, but to be honest right now I don't care so much about that, I think it would be fine to just have client-measurement only support at least to begin with!

https://github.com/felixge/sqlbench/blob/a7422d928ac6fd10ee4e91e52619da3be738ca0e/query_duration.go#L76

## Some options

I guess the "easy" option is to add another command-line flag where the user can specify a driver name if they don't want `pgx`. This is probably good enough for MVP - anyone using this tool at this stage is probably happy to get their hands dirty!

Another option could be to deduce the driver name based on the schema in the connection string URL. I'm used to this from my days of Rails, but come to think of it, I've never seen it at $dayjob for MS SQL Server.. so maybe it wouldn't be obvious, and we'd have to document what schemas are available 🤔

Any thoughts? I'm inclined to just go with the easy option and not over-engineer this thing and lock this project into unhelpful complexity on day 2 😂