Essentially, blocks can be scattered among volumes, say you have a file with 5 blocks of 1 MiB each and a 4GiB volume size. Ideally, all 5 blocks are in the same volume, so restoring needs to download 4GiB to restore 5 MiB for a partial restore.
But, worst case is that all blocks are scattered in different volumes, causing 4GiB*5 = 20GiB of files to be downloaded to restore the same 5MiB.
Also, for the new restore flow, it tries to re-use the downloaded files, meaning that they stick around in the temporary storage for longer, so you may either need a big temporary storage or have repeated downloads of the same (large) files.
