Balancer withdrawal fuzzing test

This is the main PR for creating a Balancer Metapool strategy. And when withdrawing from the strategy, the strategy contract withdraws funds from the Balancer Pool. We've identified an issue where the Balancer pool would under-deliver the funds requested by the exit pool request.

For that reason we leave the exit pool's request for min amount of tokens empty and only specify min amounts of tokens in the encoded user data request. As described in the linked issue above, if min expected token arrays on withdrawal are populated in both places the transaction reverts in some cases.

To mitigate the issue further, we also overshoot the required withdrawal amounts by 2 wei so that 1 WEI rounding error from Balancer doesn't revert the strategy's withdrawal transaction.

We would like to confirm that withdrawing rETH and stETH with different values don't cause the withdraw transaction to reverts. Withdrawing stETH could potentially be another source of problems since we are withdrawing wstETH from the pool and unwrapping it. Multiple rounding errors (Balancer pool + unwrapping) could cause larger than 2 WEI rounding error and revert the transaction.

We have some fork tests for withdrawing from wstETH pool and rETH pool. These are tests with hand-picked values, and if we could convert those to proper fuzzing tests we'd gain a lot greater amount of certainty that no combination of values would cause our transactions to revert.

Start with the existing fork tests in contracts/test/strategies/balancerMetaStablePool.fork-test.js, especially the wstETH withdrawal around line 594 and the rETH withdrawal around line 186. Replace the hand-picked withdrawal values with fuzzing coverage for differing rETH and stETH amounts, including wstETH unwrapping. Done means the withdrawal transactions do not revert across the tested value combinations.