core: gzip before sending the os image to reduce frequency of timeouts

[rcooke-warwick]

balena-os/leviathan-worker#101 - corresponding worker PR with context
Change-type: patch

[vipulgupta2048]

Instead of adding another OS dependency I was thinking if the entire problem is gzipping in the request causes a timeout then why not use the same logic to gzip out of the request something like

Create a temp file

pipeline(
  fs.createReadStream(imagePath),
  createGzip({ level: 6 }),
  tempFile
)
                         
await move(tempfile, imagePath) // https://stackoverflow.com/a/34499859 

This should also unlink/overrwite the image on imagePath