Load test: x/leverage supply and unsupply

[robert-zaremba]

Test scenario

We use the genesis file from: #1206

• create 300 threads
• in each thread
  • take an unused account from the pool (a1...a1000) -- we can't use same account in 2 threads at the same time, because it will create a nonce conflict.
  • draw a number in [0,1] of action = {0: supply, 1: unsupply}
  • perform action with 1 random token (eg 1umee)
  • put an account back to the pool
• Do it 10k times.
• NOTE: some transaction will fail because it may happen that we will use one account in a row to do more unsupply than he previously supplied, or do more supply than his available balance.

---

For Admin Use

• Not duplicate issue
• Appropriate labels applied
• Appropriate contributors tagged
• Contributor assigned/self-assigned

[toteki]

Since the qa repository has query support too (including bank balances) we could modify this to only supply a subset of token balance, or withdraw a subset of uToken balance, to prevent most transactions from failing.

The random approach would also be fine (to include more failed transactions)

[robert-zaremba]

I think it is an extra feature to have transactions which should fail. We can even predict if the tx will fail or not by calculating the balance in python, and double check the final balance at the end.