Support Redis for persistence

[fracek]

Is your feature request related to a problem? Please describe.

At the moment the only two options for persistence are:

• local file: great for development, not feasible in production.
• etcd: strong guarantees, operation nightmare.

We need a third way.

Describe the solution you'd like

Redis sits in the middle between the two options:

• it's possible to implement a distributed lock with it
• data is persisted
• many managed redis offerings on the market
• no need to define a schema. simply point the sink to it and run

The idea is to serialize the sink state to redis using a key like apibara:sink:${sinkId}. Unlike what we do in etcd, we should serialize the state as json so it's easier to debug when things go wrong.

Additional context

This blog article contains a simple implementation of a distributed lock with a single redis instance.

[fracek]

To be honest the lock method in the persistence interface can come later, it's not necessary now.