Make query execution timing more convenient to collect

On node-sqlite3 one collects query timing using .on('profile'). It's a very convenient method for collecting such data. I recently created a tool for collecting and analyzing queries on SQLITE3 using this API interface: https://www.npmjs.com/package/sqlite3-query-log

This was created while studying how to improve the performance of my application.

I'm currently evaluating whether to use libsql instead of node-sqlite3, and like what I see.

One thing I came across is that you added a feature to collect timing for run and get methods:

https://github.com/tursodatabase/libsql-js/pull/154
https://github.com/tursodatabase/libsql-js/issues/122

Going by the test cases (https://github.com/tursodatabase/libsql-js/pull/156) it appears this feature added a duration field to the object returned from run and get.

That's useful, but having written a tool for collecting and analyzing such data, my opinion is that this is an inconvenient way of collecting this data.

The use case is that a developer hoping to analyze SQL performance in their application needs a simple way to turn on and collect such data.

The approach you developed requires hunting down all run and get method calls to collect data. Further, it misses all, iterate, etc.

Start by reviewing the existing timing behavior for the run and get methods, along with the linked issues and pull requests, then compare it with the node-sqlite3 profile event API. The work is complete when there is a clearly defined, opt-in way to collect timing for run, get, all, iterate, and other query methods without locating each call site.