Prune database without finality

[michaelsproul]

Description

As per this comment: #1782 (comment), we could consider pruning minority forks from the database during periods of non-finality.

This issue exists to track this specific disk usage optimisation, as opposed to #1782 which was more general.

I'm marking it as A1 for now, as this optimisation is very contextual and not required for mainnet per se

[Ifechukwudaniel]

Hey can i do this

[michaelsproul]

Hey @Ifechukwudaniel we're working on a major overhaul of the DB in this branch: #3206. Our current thinking is that pruning the database without finality is not really necessary with the disk savings provided by the new schema.

@potuz from Prysm raised an optimisation where we could avoid importing blocks with old parents, which would help this issue indirectly and also save on lots of processing time. To make this optimisation safe, we'd have to import blocks if we see any descendant block building on them.