Count queries are expensive

[marxide]

The web interface executes several queries to display some counts. On the index page, these include the number of runs, images, measurements, and sources. Similar counts are run on other pages too, such as the run detail page.

Given the size of our database, these queries can be expensive to run. The index page count queries can take longer than a minute to complete before serving the page to the user.

These counts are not important – they're just nice to have. We should remove or optimize them. One easy optimization would be to calculate the counts at the end of a run and store them in the Run model. Then, to display the counts, we only need to query for the stored counts rather than count the relevant rows in the database tables.

[marxide]

Some of the counts are already stored in the Run model, but summing the measurement counts for all Run objects will count many measurements more than once as measurements can belong to more than one run.