Curl::download and MultiCurl::addDownload uses different logic to download files.
And judging by the codes I would say Curl::download is much much better. Why not implement/use the same logic in MultiCurl::addDownload ?

Also, Curl::download and MultiCurl::addDownload both honor cURL timeout, I had a hard time debugging why I could not download a 300mb files. I think something could also be done there.