Add a script that checks consistency between different fetching methods

[fleupold]

Given the consistency errors we had with the viewers' orderbook functions (e.g #672) this PR is creating a consistency checking script, that compares the results of the BatchExchangeViewer to the traditional (client-filtered) way of fetching orders.

The goal it to run this script against mainnet/rinkeby and get alerted when the responsed don't match.

In order to get consistent results, we need to pin the query on a specify block number (e.g. the last block of when the batch was still open to receive orders for getOpenOrderbookPaginated and the first block of when the batch was ready to be solved getFinalizedOrderbookPaginated + legacy approach).

Note, that there is still the small chance of inconsistencies if a solution was submitted in the very first block of the batch being open for solutions (this is very unlikely to happen).

Test Plan

Deploy this script in staging and let it run for the next few hours/days.

[nlordell]

Awesome!

[nlordell]

Any reason you chose to write this in JS? I'm assuming its related to missing truffle typings?

[nlordell]

ubernit: missing newline at end of file.

[bh2smith]

Note, that there is still the small chance of inconsistencies if a solution was submitted in the very first block of the batch being open for solutions (this is very unlikely to happen).

I still don't understand why not query from the block immediately before the first one published since the batch closed. This would have all the information and should have no problems of this type.

[bh2smith]

This looks very nice.

[nlordell]

I still don't understand why not query from the block immediately before the first one published since the batch closed.

I believe that has some issues with how the contract calculates balances. Don't quote me on this, as I'm not sure, but I think that the batch ID is used to determine if a pending flux should be credited to the balance or not.