Less clones & consistent search result indexes

[hasezoey]

This PR mainly focuses on doing less clones, but also does some other slight fixes that would be annoying to put in their separate PRs:

• dont clone where not necessary
• change some functions to not need to clone themself (ie dont take a reference which is always cloned)
• change all search results to start numbering at 1
• change all search indexes to be usize to prevent early wrapping (rust defaults to i32 for some reason)
• dont do some things unless in the error case (replace context with with_context)
• for first db results, dont allocate a full vector, take the first element from the iterator directly
• change PCMsg::Error to not have a duplicate url(String) as feed already contains it