We could use an aligned buffer pool [here](https://github.com/balena-io-modules/etcher-sdk/blob/70d0859f18ba72afab33a28a37e0a6fae6b22c38/lib/source-destination/block-device.ts#L221) instead of allocating a new aligned buffer every time. What do you think @mcraa @builder555 ?