SQL query logging

[david-crespo]

This may be more of a "how do I do this?" than a request for work to be done, but when I'm running the tests and debugging queries, I have to do a lot of manual intervention with println!s to see the actual SQL being executed. Cockroach has a cluster-wide logging setting, but I couldn't get it to work. (I did try turning of stdio redirection too.) This will also be useful while we're feeling out the SQL that Diesel generates.

[davepacheco]

I think this is a great idea, whether or not we get Cockroach logging to work. I was figuring we'd do trace-level slog messages, but IIRC the slog::Logger isn't passed where it needs to be today.

Other things we might consider:

• slog-tracing (@smklein have you looked at this?)
• DTrace probes for slog events so we can get lower-level log events with DTrace without restarting the program
• have an object per API request (or other operation) where we store the SQL queries run as part of that request so that we can dump them all at once at the end of the request, possibly with extra context

[bnaecker]

I happened to be going through some older issues, and think this can be closed. All the queries executed by a running Nexus can be see with ./tools/dtrace/trace-db-queries.d -p $(pgrep nexus). This was added in #488.