High-volume fuzz testing

[jalextowle]

Getting from 95% confidence to 99.99% confidence in Hyperdrive's accounting logic is a group effort among the protocol team, auditors, and formal verification experts. One of the best ways for us to contribute to this effort is building a fuzzing system that will run significant volume through an internal Hyperdrive testnet, so that we can simulate mainnet-like volumes and monitor the testnet for failures.

This initial task should involve writing MVP fuzzing bots, adding them to our infrastructure (most likely in the form of a docker-compose file), and testing them on an isolated internal testnet. We probably want an environment that is dedicated to these fuzz-bots to ensure that they don't interfere with testing of the trading bots or the Hyperdrive UI.

[jalextowle]

Some invariants that we should fuzz include:

• total shares = share reserves + shorts outstanding/c + _governanceFeesAccrued + _withdrawalPool.proceeds + epsilon
• the system should always be solvent
• trades are never executed with negative interest rates
• longs and shorts should always be closed at maturity for the correct amounts (this is easy to check)

[jalextowle]

Another invariant to include is that "negative interest mode" should never be switched on if the variable rate is non-negative. This will ensure that we're using a safe tolerance.

[jrhea]

Now tracking all system invariants in #534

[jalextowle]

Closing this since this is covered by other issues in the agent0 repo.