Given that both these functions are only called once, and are already part of the db-queries crate, I condensed them into saga_recovery.rs.

This made it marginally easier to specify arguments, and means we can re-use the same pool connection for each batch we access.

This was mostly for a minor convenience, but I can keep the old structure if you had plans to actually call these functions from elsewhere.

I think this is what the old code was doing but I don't think it's right? It may not matter. But if we get no rows here, that's not an error at all.

I added a test for this case, I think the "no node, no saga in db" case still returns a vec of length zero, so I'm pretty sure this is dead code. Updated to ErrorHandler::Server to make this slightly more apparent.

Just to be super sure, I did a quick check and it looks to me like the ErrorHandler::NotFound* variants only do anything special if the returned Diesel error variant is itself NotFound, which only happens if we use first() (first_async() for us) or get_result() (get_result_async() for us). We're using load_async(), and the NotFound docs explicitly say load() doesn't consider it an error to get 0 rows. So that explains why it didn't matter and yes I think the update you made makes sense. Thanks!