Forge block timestamps in manual seal mode

[JoshOrndorff]

This PR introduces a FAKE timestamp inherent data provider that will always include timestamps that are one aura-slot-duration apart from one another. This allows us to use manual seal to author blocks quickly without violating timestamp assumptions made by either the Aura pallet or the Timestamp pallet.

This code was taken from polkadot-evm/frontier#170. When moonbeam updates to a Frontier version that includes that PR, we should re-evaluate whether it makes sense to keep this here.

[crystalin]

Can you also try to readd the ripemd one too ? It was already fixed I believe

[JoshOrndorff]

I've tested this manually, and this change does allow manual seal to author blocks in quick succession as we expected. But it does not fix the failing tests in test-block.ts. When and why were those tests first skipped? They are working on the Frontier template, so it can't be about block production being throttled.

[crystalin]

@JoshOrndorff. My understanding is that those tests are fine with current implementatiuon of frontier + manual sealing.
However the change of Telmo + manual sealing => failing

[JoshOrndorff]

Since manual seal was not the root cause of the failing tests, I no longer think it makes sense to merge this code in here. If we ultimately want it, we should wait to get it from Frontier where it has already been merged. I'll close this and re-focus on fixing the tests.

[JoshOrndorff]

After discussion with @crystalin I understand we want this quick-authoring functionality in general, not just for one specific issue we were facing, so this should be re-opened.

I'll see if it needs updated and then get it ready for merge.

[crystalin]

Do you think we could easily remove the Timestamp + Aura requirement in the Runtime ? and do you think it is worth trying to do so (to reduce the chance of having unexpected behaviors in the tests.)

[JoshOrndorff]

To summarize our discussion in slack:

The main objective is to support both of the following in the standalone node:

• A manual seal mode that can author blocks in quick succession
• An aura mode that supports real standalone testnets

We cannot simply remove aura as it wouldn't support the second criteria.
We cannot remove aura using conditional compilation because cargo only supports one feature set per workspace.

I recommend we merge this PR as it supports both of the listed requirements and follows the same path that Frontier is following upstream cite.

[JoshOrndorff]

Tests are failing atm because the actual fix for them is in #82.

Once that is merged, we can update this and CI will be green.