fix: fix accountCreated handling, add storageTrace, fix DELEGATECALL/STATICCALL caller state

[noel2004]

TODO：update zkevm-circuits

As discussed in the issue. Some inconsistecies are caused by unsuitable initialization of stateDB used by busmapping. This PR change the initialization by adding states being traced in storageTrace field of block results and this work should fix most of the inconsistency.

This PR also change the example traces by manually revert the byte order in proofs, because they have been unified into big-endian.

[noel2004]

I have test this PR on the skipped blocks in roller and all of them can passed the handl-block phase (so witness can be assigned and circuit be built), execpt for block 17, which assert with !callee_code_hash.is_zero().

[icemelon]

@noel2004 I checked the transaction in the block 17 and it calls to the L2Messenger contract. The contract shouldn't have a zero codehash. So not sure why the assert failed for this block.

[lispc]

You can Leave this error case to me. You can really proved some blocks in our dev server (i suggest only prove evm circuit. Don't prove agg)

[lispc]

Suggest Set MOCK_PROVE env var to true to get better circuit error msg.

[0xmountaintop]

I will build and test without merging first.

BTW what dores issue.json mean?

1. it's not block 17.
2. issue inconsistency between state rebuilding #36? but there're 4 problems there.

[noel2004]

I will build and test without merging first.

BTW what dores issue.json mean?

1. it's not block 17.
2. issue inconsistency between state rebuilding #36? but there're 4 problems there.

issue.json is being added accidentally. It would be prune

[noel2004]

Now the PR has passed with block 17 so it should be ready.

Work left:

• lacking of serialize in storeTrace may cause roller fail for its store / reloading process.

[noel2004]

Now the blockResult structure can be fully serialized / deserialized.