hypothesis.strategies.integers() is inclusive, but microseconds values
greater or equal to 1 million are rejected by the model

This allows callers to pass different (typically, stricter) strategies
to hypothesis instead of filtering the output; which makes example
generation faster.

It will be used by swh.storage.backfiller (so indirectly, swh.scrubber)
to load directories from the postgresql database, whose schema accidentally
allowed directories with duplicate entries -- without corrupting the
shape of the directory too much.

We can use `format_git_object_from_parts` inside it.

This would be useful for the IPFS bridge, and seems good to complete the
API in any sense.

it's a piece of information used several times in the swh stack.

Related to T3922

Hypothesis: This makes the documentation build fail.

This changes the deprecated :show-nested: instruction to the :nested: one [1].
This also fixes warning about misdefined block [2]

[1] https://sphinx-click.readthedocs.io/en/latest/usage/#directive-click

[2]
```
10:38:38  Warning, treated as error:
10:38:38  /var/lib/jenkins/workspace/DMOD/tests-on-diff/docs/cli.rst:14:Literal block expected; none found.
10:38:38  make: *** [../../swh-docs/Makefile.sphinx:32: sphinx/html] Error 2
10:38:38  make: Leaving directory '/var/lib/jenkins/workspace/DMOD/tests-on-diff/docs'
10:38:38  ERROR: InvocationError for command '/usr/bin/make -I ../.tox/sphinx/src/swh-docs/swh/ -C docs' (exited with code 2)
```

committer=None happens on some malformed commits generated by old dgit
versions ; and it is possible for author=None to happen for the same reason.

these fields are computed attributes and may be removed from the backend
storage. Helps writing tests.

This will be used by swh.storage.backfill's tests.

This reverts commit a0f54362.

This means this commit removes the 'offset' and 'negative_utc'
arguments.

It also removes the 'negative_utc' attribute (which is not used
anymore), but keeps an 'offset' property, which is an alias to
'offset_minutes()'. This is only to keep this commit readable;
the next commit will remove this alias.

This should slightly reduce the memory used by dir/rev/rel objects

A future commit will remove the 'offset' attribute, and rename
'offset_bytes' to 'offset'.

Modules using the `.offset` attribute will need to switch to this new
method to keep working.

This allows keeping compatibility with existing users of the TimestampWithTimezone constructor

'offset' becomes a property instead of an attribute and constructor argument.

This also removes both from the output of `.to_dict()`.

This is step 6 of https://forge.softwareheritage.org/T3752

This will break packages that still use the constructor directly,
ie. swh-storage and swh-loader-git (and tests of swh-loader-core
and swh-loader-svn)

This is needed when recomputing the manifest of git objects with weird offsets.

blake2s and blake2b have been provided by the stdlib hashlib since
Python 3.6, and we declare 3.7 as minimum Python version supported.

1. Most objects do not need it so it's a waste of space
2. This means we just extend the existing format (some objects will
   have that key in their dict) instead of changing it (retroactively
   adding it to all objects)

This will be used to store the original manifest of 'weird' git objects,
when we cannot reasonably represent them otherwise.

Using .now() produces data that differs between xdist processes,
as files are imported after forking, and xdist requires consistent
data across processes.

It calls attr.validate() (which calls the validators), and recomputes
the hash of HashableObject instances.

A future commit will also make it check the raw_manifest attribute
when relevant

It's just simpler this way

For the sake of completeness (a future commit may depend on it).

For now it is filled from 'offset' and 'negative_utc', but it will replace them
in a future commit.

This is to simplify and add support for more 'weird' offsets we do not currently
support.