Update to upstream version '2.6.4'
with Debian dir 92f45f224d492f3f0b914e640c27ce17d2a9dcb7

We agreed a while ago they should be IRIs and not just URIs.

This will trigger crashes in swh.storage.cassandra, as
currently expects (wrongly) that origin urls are ASCII.

* empty fetcher name or version is not accepted by cassandra
  (and is nonsensical anyway)
* ditto for non-ASCII (and any non-printable is nonsensical)
* null bytes/chars are accepted by neither postgresql or cassandra

Update to upstream version '2.6.3'
with Debian dir 216fcda7a93684ee59ff01a0c7f8fc6021ad0e72

This is the only value we should use from now on;
and the value is ignored by swh-storage anyway.

Update to upstream version '2.6.2'
with Debian dir df6eb8e5ae3566559bfc86ac892acd3504dc130b

It will be used by swh-web.

Closes T3393

Update to upstream version '2.6.1'
with Debian dir de4cc1005c62c873540e9c8b8067233771377a1f

- add typing annotation to avoid such error in the future

Fixes T3383

Update to upstream version '2.6.0'
with Debian dir a5928cf48b5ed5e74f56196c93fd182d82f7589a

the problem was for datetime<epoch, the timestamp is negative, but since
it's a float that includes the microseconds, if both are true (< epoch
and microsecond > 0), then the computed (int) timestamp was off by one.

Add dedicated tests for this.

Update to upstream version '2.5.1'
with Debian dir 7de8b2d4d70f1a04bd64715de0a11f2f538c8fe1

make sure the snapshot id in OriginVisitStatus refers to existing
Snapshot objects.

Update to upstream version '2.5.0'
with Debian dir 05035d9eb3380a44d4daa228462b7349d05b90b7

If run after py3-full, it overwrites the full coverage info, so most lines
are incorrectly reported as uncovered in Phabricator diffs.

The git_object is what will be actually useful to the vault.

It's also easier to test, because test_identifier.py has the
entire git_object in its test data.

Before this commit, manifests were only computed internally before
hashing, so they were not available to outside modules.

This makes testing the module very painful, because identifier
functions can only be tested by checking the hash; so test failures
did not show mismatches between the computed manifest and
the expected one.

Additionally, the 'git bare cooker' of the vault is likely to use
these as well, as it needs to format git objects in the same format.

Update to upstream version '2.4.2'
with Debian dir 196d5f227c50e9b78022da184ababc6ee414f976

There is a regression that breaks attr.evolve() when updating attributes
that contain an attr class; which we use (eg. for Person or TimestampWithTimezone).

v21.2.0 is expected to fix the issue, but won't be released immediately:
https://github.com/python-attrs/attrs/issues/804#issuecomment-833471190