Handling of genesis block takes too much time

[ManuGowda]

Expected behavior

Even with a large genesis block file, it should be able to sync the nodes properly.

Actual behavior

• Procedure getHighestCommonBlock for given block ids returns the whole common block header, in case of testnet or mainnet genesis block header the size of genesis block header is too big (20-200MB).
• When getting the block, it tries to decode the asset which takes long.
• Genesis block are kept in memory

Steps to reproduce

• Migrate 2.1.6 and generate genesis block (which is around 200MB)
• Run lisk-core start -n mainnet with the new genesis block
• Connect to running node as seed peer
• Observer the syncing node fails to request block during synchronisation

Which version(s) does this affect? (Environment, OS, etc...)

5.1.1, 3.0.0

[shuse2]

General direction of solving this issue will be

• lisk-chain package will not decode the block header asset, and it will be encode/decode only when it's needed. (ie: for hooks etc)
• Add genesis data to consensus state (init round and height) so that genesis header is not needed in the memory
• Update getHighestCommonBlock to return only ID of the common block