Putting large texts into Attachment.content multiple times might require lots of memory.

Proposal: have an option to allow Attachment.content to be streams/buffers, that are only read/processed when it comes to normalizing the data?

Attention: check compatibility for browsers. For node streams and buffers exist.
But they are not part of the common JavaScript implementation ala ES.
As long as there is no clear path, this feature stays just an IDEA.

The thing is, that the Attachment.encoding and the buffer's content might cause some interference,
when the encoding claimed to be base64, but the buffer is not in binary mode, so there is no chance to fetch and convert it properly.