Update to upstream version '2.0.0'
with Debian dir e756045d5a5f2efe9eff53ef16d69d07ecc9afe4

As dulwich's client.fetch_pack expects an instance of history graph
walker with set of known heads, move the local heads caching from
`determine_wants` to the RepoRepresentation initialization logic.

Our previous code would always initialize the graph walker with an empty
set of heads (as the `graph_walker()` method is called before
`determine_wants()` has run, so `self.heads` was always empty), so we
would never actually fetch an incremental pack file.

Update to upstream version '1.10.1'
with Debian dir faa5c5d1344eabafd61cab6de6e88dd7a793d017

ShaFile.get_type was deprecated and has been removed.

New typings have been added in dulwich that trigerred a new mypy error.

'/' is very rare, but it is rejected by swh.model, because it would cause
issues in other places (eg. cannot be accessed via the web UI or web API)

To avoid any confusion, as raw_manifest contains the git header

- pre-commit from 4.1.0 to 4.3.0,
- codespell from 2.2.1 to 2.2.2,
- black from 22.3.0 to 22.10.0 and
- flake8 from 4.0.1 to 5.0.4. Also freeze flake8 dependencies.

Also change flake8's repo config to github (the gitlab mirror
being outdated).

Use helper fixture loading_task_creation_for_listed_origin_test from
swh-loader-core and remove redundant tests.

Update to upstream version '1.10.0'
with Debian dir 294d6835e809f48c638b3d1709768566a7823c76

swh.model.hashutil recently got type annotations

Dulwich 0.20.43 dropped the double caching of HTTP responses so
we can now remove comments about that issue.

Related to T4311

Update to upstream version '1.9.1'
with Debian dir 19202ded7b36bfa220f2cacd15ec2aabcf19bcf2

It does not really make sense to average percentages, because they are unweighted,
and do not tell anything more than totals about individual runs.

Update to upstream version '1.9.0'
with Debian dir e8cc8071d3eadf2f9ced9823512976a761d2e5b8

In preparation for the removal of DVCSLoader from swh.loader.core,
as the git loader is the only one to use it anymore.

Update to upstream version '1.8.1'
with Debian dir 70839ed444fd62d05f581b3bd1c92cb174933e3d

base_snapshot_reverse_branches needs to contain all objects that
may be a snapshot target that the remote did not send to us.

Because we now use all snapshots to build the "have" list, such
targets include all targets of snapshots of "parent" origins,
not just the previous snapshot of the current origin.

This typically happens when a forge-fork pull branches from its
parent.

Resolves Sentry issue [[ https://sentry.softwareheritage.org/share/issue/1c08f5d764e7494e83ba254dc47f17af/ | SWH-LOADER-GIT-102 ]]

Only instances of (subclasses of) AbstractHttpGitClient have this attribute.

For other instances, we can consider it to be False, because the dumb
protocol only exists over HTTP(S).

This issue was found by mypy, thanks to the addition of type annotations
to changes in dulwich 0.20.36 affecting mypy's type inference.

Update to upstream version '1.8.0'
with Debian dir 79f3cb38e7ce46623822871ba88305aed21b1f76

Before this commit, determine_wants() used the origin's last snapshot
if any, or the closest parent's snapshot if not.

However, we noticed that many repositories that are very slow to load
are forks that were already visited, but their owner rebased it on the
parent since the last visit, causing potentially many commits to be
added to the origin.

This ensures we do not needlessly fetch these new commits when we
already loaded the parent.