I fell down a bit of a rabbit hole while exploring for performance improvements for #6370.

I made the following improvements:

• changed queryStruct to use tx.Get instead of queryFunc
• fixed a latent bug in the table joining code that would add the join's args to the args list even if the join was not added
• omitted joining tables in the get aggregate query that were necessary only for ordering. Minor increase in performance when getting object counts, total file sizes, etc.
• made all calls to the sqlite interface pass the context

The last item was the most important one. This fixes what I believe is a pretty nasty latent bug, where queries would not be canceled when the applicable http request was canceled.

An an example, on a database with 32m images - where each image request takes around a minute - I flicked through pages, changed the sort order and direction, and the page sizes all before any results came back. I had to force kill the process because it was cooking my server 😬

A nice side-effect of this is that you can now kill the stash instance while a query is running if you cancel the request via navigating to another page in the UI (assuming the request came from the UI). Previously it would hang until the queries were complete - unless you kill -9 it.