improve delayedWeth test and coverage

[AmadiMichael]

I'm of the opinion that the withdraw(uint256) method used in DelayedWETH tests should be that of the DelayedWETH contract (which in turn calls the withdraw(address,uint256) method). SInce the former calls the latter, that means the latter is tested for too indirectly.
This helps with test coverage.

Other alternatives:

• Write tests for withdraw(uint256) also which (the test) are basically a clone of that of withdraw(address,uint256)? This is also a better approach maybe since modifications to any of the 2 functions will be tested for already.
• Leave it as is?