perf(Filters) - increase performance of updating charge filters

[nudded]

Description

Based on profiling information we found out that the to_h method on ChargeFilter is called often when updating a charge with a large ( > 100) amount of filters.

A second observation is that the touch method means we're updating all charge filters every time, which has poor performance when you have a lot of those for a given charge.

Changes

• cache to_h
• updated_at is used for ordering the filters, but if you have > 100 we've chosen to optimize and not touch all records. This means stuff will change in the UI, but at that amount nobody is really managing them based on order.