Improve error handling for remote resources

Error handling for remote resources (e.g., SPARQL queries, but this also applies to HTTP resources) should be improved. Right now, most HTTP error codes will just panic – we should turn this into a nice error message.

Start by tracing the remote-resource handling described in the issue, including SPARQL queries and HTTP resources, to find where HTTP status errors become panics. Done means typical HTTP error codes produce clear error messages instead of panicking; no specific files or tests are named in the payload.

Written by the indexing model from the issue text.