to prevent confusion with setup()'s "extras_require" argument.

related to T1242

it makes life a bit easier for CI (less dependencies).

This is a temporary move.  This will eventually converge with
swh.model.cli as the directory hashing ability overlaps.  What's
missing in the cli for now is the ability to filter patterns.

This is mostly used to compute hashes on tree coming from multiple
dvcs. This imposes the need to be able to filter pattern like .svn,
.hg, etc...

Related D454

This aligns the new api with the old one.

Remove unused track_length parameter

And revert to their original behavior (returning only digest as bytes)

This allows calls like:

- MultiHash.from_file(file_object).digest()
- MultiHash.from_path(b'foo').hexdigest()
- MultiHash.from_data(b'foo').bytehexdigest()

This is the first step to improve the hashutil module according to
"the" plan [1].

In this regards, the hashutil exposes the same function as before.
Internally though, they now uses a MultiHash instance.

Related D410

[1] D410#7952

Related T421

Related T421

Related T1152

... and add an option to explicitly not follow them, if needed.

Closes T1135

Closes T1134

with explicit option to hide it

Closes T1133

This commit changes the swh pids representation returned by the
parse_persistent_identifier function from a dict to a named tuple.
Also, the 'object_type' attribute of that named tuple now contains
a full name instead of an abbreviated one.

Closes T1112
Closes T1125

Close T1104
Close D346

Related T1104

Related T1104

As the client of this api might not be aware of what's the right id
key for a given object.

Related T1104

Related T1104

Related T1104