feat(search/cache): use cache to remove duplicate queries

[ShanaryS]

cross-seed search will now never perform duplicate queries. After searchee creation, the torznab search is calculated for each and those with the same are grouped back to back. When searching, the IndexerCandidates from the last search is cached and if it applies to the current search will be used (the map will ALWAYS have a size of 1, indexers not searched by previous are searched and added to cache). This allows duplicate queries to only hit each indexer once (per cross-seed search) and to be assessed near each other. This is mainly useful if you have multiple variations of a torrent like: different groups, resolution, source, etc which are not duplicates but will have duplicate queries. From my testing, my queries were cut in half.

These grouped searchees are ordered by most files first, then if .torrent based. We prefer most files first as that reduces the chance of a MATCH_PARTIAL happening when another searchee could have been a full match. I also changed how dupes are defined. Rather than just name, I check within the grouped searchees and find ones with the same size and number of files, preferring torrent based.

A nice benefit of the implementation is that if the previous searchee needed to search an indexer than the next wasn't going to, the results are still used by the next. Meaning that whenever we search for something, all searchees that care about the query will update their decisions, effectively ignoring excludeRecentSearch and excludeOlder if we are making the search anyways. This won't be the case initially but will eventually be true once a excludeRecentSearch cycle completes.

---

Updated some defaults in the config. Set snatchTimeout to 30s and searchTimeout to 2m. I think this is much better than leaving it at infinite. I also changed back maxDataDepth to 2. With the new caching, there is no risk of duplicate queries. And while partial reduces the need to go above a value of 1, in the cases of multi season releases 2 will be useful.