Adding abstractions for the omni-account in the worker

[silva-fj]

This PR introduces some abstractions related to the omni-account in the worker.

The main changes in this PR are:

• New primitive types (related to omni-account).
• Litentry core abstractions: OmniAccountRepository is an abstraction to request data from the parachain and InMemoryStore is an abstraction to store and manage the OmniAccount Members in RAM.
• Optional header parameter to get_storage_keys.

TODO:

• Replace usage of state key repository with the new key repository (aes256 key) for the on-chain data encryption.

[linear]

P-1086 Add abstractions for omni-account in the worker

[silva-fj]

Thanks!

I'm a bit confused about all the changes in StfExecutor, why would we need those? I naively assume in the in-RAM AccoutStore has nothing to do with sidechain state (thus stf), so mutation of it doesn't need to go through the stf executor. Is it the case?

The idea was that when handling a trusted call, we are able to:

1. Query the latest state of the AccountStore to be sure we have the latest hash before sending the extrinsic to mutate it. (We could rely on the InMemoryStore for this), this is the reason for injecting the ocall_api and the parentchain header.
2. We need to be able to encrypt the data before sending the extrinsics and decrypt the data when handling trusted getters. So we inject the encryption key repository.

Does it makes sense? Do you had another approach in mind?

[silva-fj]

Another approach could be:

• We keep the data unencrypted in the InMemoryStore, so we won't need to decrypt it for the trusted getters.
• We rely on the InMemoryStore to get the current AccountStore and hash when handling the trusted calls.
• We create a channel/thread that would be responsible for encrypting the data and sending the extrinsics to the parachain. So when handling the trusted call we verify and send the message to this channel. The problem with this is that we won't be able to wait until it's included in the block.

This way we won't need to inject anything into the executor. What do you guys think? Any other ideas? @Kailai-Wang @kziemianek