Fix bug in fraud proof transaction fee value, and combine storage

[teor2345]

The fraud proof was calculating the consensus transaction byte fee as 1 when dynamic cost of storage was disabled. But the consensus_chain_byte_fee() runtime API returned DOMAIN_STORAGE_FEE_MULTIPLIER in that case. This PR makes the values consistent by fixing the fraud proof calculation.

It also moves the fee into the combined fraud proof storage.

Part of ticket #3281.

Code contributor checklist:

• I have read, understood and followed contributing guide

[NingLin-P]

Good catch!

To further avoid such inconsistency I think we can use InvalidInherentExtrinsicData (perhaps need rename to InherentExtrinsicData) to construct the inherent extrinsic in domain CreateInherentDataProvider instead of querying these data through consensus runtime API one by one.

[vedhavyas]

make sense!

Refactor seems to be worth it!
Good to go once Ning's suggestion is included

[teor2345]

To further avoid such inconsistency I think we can use InvalidInherentExtrinsicData (perhaps need rename to InherentExtrinsicData) to construct the inherent extrinsic in domain CreateInherentDataProvider instead of querying these data through consensus runtime API one by one.

I’d prefer to make this change after all these PRs merge, because it would make each of my 3 open PRs more complicated. It seems like a cleanup we could do once all the fraud proofs are combined.

The base branch was changed.