`lineReader` and `function transform` buffer the entire file before writing and this will make strategies run out of memory

[TimDaub]

• consider that we have block log files that can get up to 400GB in size after the extraction process
• But currently how we're implementing lineReader and function transform in the lifecycle is that a buffer accumulates the entire file in memory: https://github.com/neume-network/strategies/blob/3b0480d9c5e0a39e3f87cea410fc19207f788794/src/lifecycle.mjs#L27
• And then we write the file after completing the transformation strategy: https://github.com/neume-network/strategies/blob/3b0480d9c5e0a39e3f87cea410fc19207f788794/src/lifecycle.mjs#L88
• This will inevitably lead to memory overload especially with 500GB large files
• A solution would be where we write continuously in an append-only fashion to the file