Extensive OOGX tests involving memory expansion + gas estimation tool BytecodeRunner.runOnlyForGasCost + MSIZE (#1783)

[lorenzogentile404]

Closes #1685

• BytecodeRunner.runOnlyForGasCost Method to calculate cost dynamically - based on LondonGasCalculator (Besu)
• OOGX for opcodes with no mem expansion : list from opCodeToOpCodeDataMap.values() without opcodes with mem expansion cost
• OOGX for opcodes with mem expansion cost : MSTORE, MSTORE8, MLOAD, CALLDATACOPY, CODECOPY, EXTCODECOPY, RETURN, RETURNDATACOPY, REVERT, CREATE, LOGX
• MSIZE test: with non zero value and post memory expansion

Local benchmark done on 366 tests with

• calculate gas cost manually 1min40s
• calculate gas cost dynamically 1min30s
• calculate gas cost by running BytecodeRunner.run twice 2min50s

Test coverage improvement
+7 lines covered
+4 conditions covered
Results
+0.3% condition coverage on Unit tests
+0.1% overall coverage score (with blockchain ref tests)

[OlivierBBB]

Obvious remark: both MSTORE's expand memory, the first to size 0x20, the second to size 0x40.

[amkCha]

You're right, it felt better to go from a non-zero value to a non-zero value, but it's strictly the same behavior. Let me simplify it 👍

[OlivierBBB]

This paragraph could be extracted into a method I imagine ? It's duplicated quite a number of times.

[amkCha]

Good idea, I created a Utils method 👌

[OlivierBBB]

Here, too, we have the opportunity to test simultaneous exceptions, in this case:

jumpException
outOfGasException

And it would be trivial to implement, just switch the second push to

    final Bytes bytecode =
        BytecodeCompiler.newProgram()
            .push(1) // pc = 0, 1
            .push(6) // pc = 2, 3
            // ...

[OlivierBBB]

It would make sense to create a new class for these MultiExceptionTests.

[OlivierBBB]

Which would include the RETURNDATACOPY double exception test mentioned earlier.

[amkCha]

Issue created for MultiExceptionTests #1838

[OlivierBBB]

Left several comments

[OlivierBBB]

LGTM