Node.js — Compress a File with pipeline()

Rotated logs (an earlier exercise) should be compressed: text logs shrink 10× with gzip. Node's
[`zlib`](https://nodejs.org/api/zlib.html) compresses as a stream, so a 5 GB log never needs
5 GB of memory — provided the streams are connected properly, which is
[`stream.pipeline`](https://nodejs.org/api/stream.html#streampipelinesource-transforms-destination-options)'s job.

Your task

Write async archive(src, dest):

1. Stream src through zlib.createGzip() into dest + '.partial' with
pipeline from node:stream/promises.
2. On success, rename the .partial file to dest (replacing an older archive) and return
{ bytesIn, bytesOut } — the sizes of src and dest in bytes.
3. On failure (missing source, source is a folder...), remove the .partial file and re-throw
the error. An existing dest must be left untouched.

The driver checks the sizes, decompresses dest to compare it with the source, and lists the
folder afterwards: no .partial file may remain.

Why not .pipe()?

a.pipe(b).pipe(c) does not forward errors: if a fails, c stays open forever and nobody
is told. pipeline destroys every stream on the first error and reports it once.