refactor: Denormalize action_receipt_actions table in order to improve common queries performance

[khorolets]

Closes #87

Starting to address the issue. Plan:

• [x] Migrations to add columns and fill them with the data
• [x] Update corresponding model (ActionReceiptAction)
• [x] Update corresponding db adapter which stores ActionReceipt actions to include data for newly added columns
• [x] Adjust adapter to extend args field to include args_base64 (is there from the start) and args_json if possible (in case of successful deserialization)
• [x] (considerable) Add indexes on args->>'args_json'->>'receiver_id' and args->>'args_json'->>'amount'
• [x] create migration script/command to iterate over all function call args and try to parse args_json to fill historical data along with newer data

[khorolets]

@frol please review, I've checked args logic on localnet and it works. I am hesitating about the new version. I've put 0.6.1 as I don't think we've changed a lot to bump to 0.7.x any objections?

[khorolets]

For history: While this PR was in progress we've released a few minor version so we end up with 0.6.3 version of Indexer for Explorer in this PR.

[khorolets]

Out data is too big and transaction-based migration queries are locking tables for huge amount of time. We end up in using custom python script splitting big query into smaller ones. Here's the script for the future use.

from sqlalchemy import create_engine

START = 1595370903490523743
STEP = 1000000000000
END = 1620717648739546794
ESTIMATED_STEPS = (END - START) / STEP
connection_string = 'postgresql+psycopg2://user:pass@host/database'

def generate_sql(from_timestamp: int, to_timestamp: int) -> str:
    """
    Generates str for SQL query to convert args_base64 to args_json if possible
    """
    return """
    UPDATE action_receipt_actions
            SET args = jsonb_set(args, '{args_json}', decode_or_null(decode(args->>'args_base64', 'base64')), true)
            WHERE action_kind = 'FUNCTION_CALL'
                AND receipt_receiver_account_id != 'client.bridge.near'
                AND receipt_included_in_block_timestamp <= %d
                AND receipt_included_in_block_timestamp >= %d;
    """ % (to_timestamp, from_timestamp)

if __name__ == '__main__':
    print("Establishing connection to %s..." % (connection_string.split('/')[-1],))
    engine = create_engine(connection_string)
    print(f"Estimated queries to execute: {ESTIMATED_STEPS}.")
    from_timestamp = START-STEP
    to_timestamp = START
    counter = 1
    with engine.connect() as con:
        while True:
            from_timestamp += STEP
            to_timestamp += STEP
            if (END - to_timestamp) < STEP:
                break
            print(f"{counter}/{ESTIMATED_STEPS} (from {from_timestamp} to {to_timestamp}")

            out = con.execute(generate_sql(from_timestamp, to_timestamp))
            print(out.rowcount)
            counter += 1

    print("FINISHED")