Clippy: Fix result_large_err by boxing redb errors

[patowen]

The ReadTransactionStillInUse error variant of the redb error type is 160 bytes long, with most of its size coming from the redb::tree_store::TableTree type.

An alternative implementation would be to ignore the error or add a "clippy.toml" to increase the error threshold to 161, but 160 bytes is quite long for a Result type, so boxing the error would likely be the better design.

Note (for anyone reading the comment thread): This PR is a revised version of a PR that initially just ignored the clippy lint in the places it triggered from.

[Ralith]

Typically you'd address this lint by boxing the error. Is there a reason we shouldn't do that here?

[patowen]

No particular reason from me. I don't have a good sense of how much we care about the error sizes, and I didn't expect that it would be a bottleneck, so I went with the easiest approach (of not changing the code). I also checked the redb examples, and they also took this shortcut (https://github.com/cberner/redb/pull/974/files).

Boxing the error does seem better though. I need to figure out how to do this in an idiomatic way that interacts with thiserror. 160 bytes is indeed very big.

EDIT: PR updated to stop ignoring errors. I'm not sure if there's a more idiomatic way of doing this.