## Algorithm
Given any digest algorithm `H` (a Hash function algorithm),
a `RecursiveDigest(H, path)` is:
* for a file: `H("F" || file_content)`
* for a symlink: `H("S" || symlink_content)`
* for a directory: `H("D" || directory_content)`
As you can see a one-letter ASCII prefix is used to make it impossible
to create a file that has the same digest as a directory,
etc. The drawback of this approach is that `RecursiveDigest(H, path)` of
a simple file is not the same as just a normal digest of it (`H(file_content)`) .
`file_content` is just the byte content of a file.
`symlink_content` is just the path the symlink is pointing to, as bytes.
`directory_content` is created by:
* sorting all entries of a directory by name, in ascending order,
using a simple byte-sequence comparison
* for all entries concatenating pairs of:
* `H(entry_name)`
* `RecursiveDigest(H, entry_path)`