processors: optimize via caching

[slint]

Some of the processors are actually run multiple times in a row for very similar sets of values, since requests/events from the same (or similar) user are happening close to each other. Some examples:

• GeoIP lookups
• Robot/Machine detection
• Any DB-dependent processors

To optimize these processors, simple functools.lru_cache decorators can be added to heavy-duty functions (this might require some function calls/signatures refactoring). On top of that, a Redis-based LRU cache can be used in order to share/persist the cache between different processes/workers/nodes.