fix(tests/benchmark): update jump seq for valid dest

[LouisTsai-Csie]

🗒️ Description

There is an issue with the original test case: the jump target is not a valid JUMPDEST operation.
More details could be found in the discussion.

The updated version precomputes valid jump destinations to improve the density of JUMP operations.

The repeated pattern is structured as follows:

0x00 PUSH2 DEST
0x03 JUMP
0x04 JUMPDEST
...

The first JUMPDEST is at offset 0x04. Since each pattern occupies 5 bytes, subsequent JUMPDEST locations can be calculated as 4 + 5 * i, where i is the pattern index. The total code size must remain within the allowed limit.

🔗 Related Issues

✅ Checklist

• All: Set appropriate labels for the changes.
• All: Added an entry to CHANGELOG.md.
• Tests: Included the type and version of evm t8n tool used to locally execute test cases: e.g., ref with commit hash or geth 1.13.1-stable-3f40e65.
• Tests: Ran mkdocs serve locally and verified the auto-generated docs for new tests in the Test Case Reference are correctly formatted.

[jochem-brouwer]

This looks good to me, but CI is failing 😅

[LouisTsai-Csie]

I've tried rebased for several times, but i will give it a try again.

[LouisTsai-Csie]

Thanks @marioevz for explanation, we've checked the reason for the failing CI.

There are two issues in the evmone coverage.

0 :         state.status = EVMC_BAD_JUMP_DESTINATION;753               
0 :         return nullptr;

This originates from the previous test case, since the code sequence will jump to an invalid destination.

0 : inline code_iterator pc(StackTop stack, ExecutionState& state, code_iterator pos) noexcept: {
0 :     stack.push(static_cast<uint64_t>(pos - state.analysis.baseline->executable_code().data()));
0 :     return pos + 1;

This also originates from the previous test, it uses PC opcode but the fixed one does not use this operation.

Since this PR intends to fix the issue in the JUMP operation, these CI error is normal here

[jochem-brouwer]

LGTM. Not sure why CI fails.