chore: storenode cycle improvements

[richard-ramos]

#1223 (review)

• Interface between status-go and waku api should be simple i.e status-go specifies list of storenodes to be used and waku internally manages connectivity to them and picks a storenode and manages active store. There can be one event emitter than status-go can register with to get all updates such as what is active-store-node, in case connectivity is lost etc.
• All 3 components i.e history-query, missing-messages and message-check should end up using same set of store nodes instead of their own selection criteria and logic.
• selection of storenode can be done using peer-selection and RTT. logic seems duplicated here
• store node failed requests should also consider requests made from missing_messages and message_check so that active storenode can be switched.

status-im/status-go#5857 (comment)

• use a state machine for different status of store node to simplify the channels

#1223 (comment)

• simplify PerformStorenodeTask