Fetch information about short-lived Codespaces

[iaindillingham]

The codespaces task (metrics/tasks/codespaces.py) fetches information about the Codespaces that exist when the codespaces task is run. Like all tasks, the codespaces task is run daily (metrics/app.json). Consequently, the codespaces task doesn't fetch information about Codespaces that are started and deleted between task runs; such Codespaces are short-lived, because they have a maximum lifespan of 24 hours.

It's important to fetch information about short-lived Codespaces, as they may represent researchers who are unable to use Codespaces effectively. Including this information in our Codespaces dashboard would allow us to identify and contact researchers, and ask them about their experience of using Codespaces.

Running all tasks more frequently is a quick, but ineffective solution: I anticipate that GitHub will prevent us from doing so. However, running some tasks on one frequency and other tasks on another frequency will involve modifying our current -- elegant! -- approach to ensuring that all tasks are run (metrics/tasks/__main__.py).

[Jongmassey]

If the "codespaces" task is moved to a different entry point and schedule, obvious the dokku cron config will need updating but also the sentry cron monitor config