Update basefee implementation to use Uint256 rather than long

[garyschulte]

Besu's implementation of EIP1559 is using a long instead of Uint256 for baseFee. This is not likely to be a problem in the near future since baseFee would have to be in excess of 9 eth to cause an overflow, but we should match the types in the spec.

Acceptance criteria:

• update implementation and tests to use Uint256 baseFee

[fab-10]

@garyschulte should not we use Wei instead of UInt256 since it is a fee?