There is more repetition, but it's easier to read and
'%timeit Revision.from_dict(d)' is 5 times faster.

to the benefit of MultiHash api

Summary: to make sphinx happy.

Test Plan: make -C docs

Reviewers: ardumont

Reviewed By: ardumont

Differential Revision: https://forge.softwareheritage.org/D571

so pytest does not think they are test classes.

to clear deprecation warnings.

related to T1261

Summary: Related to T1261

Test Plan: before and after: 138 tests passed (2 skipped)

Reviewers: #reviewers!

Differential Revision: https://forge.softwareheritage.org/D520

and ensure it is included in the source package.

Related to T1265.

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