Process genesis for initial state

[bryanchriswhite]

Acceptance Criteria

1. The repo contains a script which downloads the network Genesis from a given URL.
2. The schema includes a NativeBalanceGenesis entity.
3. The repo contains a script which processes the downloaded Genesis according to the needs of various features (i.e. entities which depend on it).
4. These scripts are executed automatically at some point in the startup of the subquery node.

Proposal

1. An environment variable seems like the most straightforward place to store the genesis URL.
2. Example:

type NativeBalanceGenesis {
  id: ID!
  account: String!
  amount: BigInt!
  denom: String!
}

3. It probably makes sense for the initial version of script to interact directly with the database. We can use the same tooling as our end-to-end tests (python + psycopg) to avoid introducing additional dependencies and keep tooling consistent. Perhaps we'll want/need to upgrade to something more robust and/or javascript-native in the future. I would also propose that it would be good for maintainability and onboarding if the script employed inversion of control to support distinct genesis state handler functions, defined separately. I would suggest considering ReactiveX (PyPI) which implements the observer pattern flavor of IoC. This design also allows for concern-scoped handlers to be reasoned about and maintained independently.
4. I imagine we can use a kubernetes job to run through once at the beginning of indexing but we should ensure that the job doesn't start until after the SubQuery node is up (i.e. has created all tables).