This avoids the confusing AwsLocalAzure and AwsAzureLocal labels

By using read-only shared access signatures for RO access, we can reduce
the potential attack surface.

This references an objstorage instance that's not being used anymore

This is backed by the IP for both hosts, so should allow accesses to
continue even when one of the hosts is down.

As the queue for those is a bit high.

It initially got stopped some time ago when the storage1 was almost full. Since then, a
new storage got installed.

Related to swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5275

Current state:

```
2024-04-16 15:05:36 softwareheritage-scheduler@albertina:5432 λ select type, count(*) as total
from task t inner join task_type tt using (type)
where t.type like 'list-%' and
  policy = 'recurring'
  group by type order by total desc;
+------------------------------+-------+
|             type             | total |
+------------------------------+-------+
| list-gitlab-full             |   185 |
| list-gitlab-incremental      |   181 |
| list-cgit                    |   133 |
| list-gitea-full              |   121 |
| list-gitweb                  |    76 |
| list-gitiles                 |    45 |
| list-stagit                  |     9 |
| list-phabricator-full        |     5 |
| list-pagure                  |     4 |
| list-gogs-full               |     4 |
| list-debian-distribution     |     4 |
| list-nixguix                 |     4 |
| list-opam                    |     2 |
| list-pypi                    |     1 |
| list-sourceforge-full        |     1 |
| list-sourceforge-incremental |     1 |
| list-golang-incremental      |     1 |
| list-bower                   |     1 |
| list-cran                    |     1 |
| list-github-full             |     1 |
| list-github-incremental      |     1 |
| list-gnu-full                |     1 |
| list-bitbucket-incremental   |     1 |
| list-launchpad-full          |     1 |
| list-launchpad-incremental   |     1 |
| list-maven-full              |     1 |
| list-npm-full                |     1 |
| list-packagist               |     1 |
| list-pubdev                  |     1 |
+------------------------------+-------+
(29 rows)

Time: 6322.594 ms (00:06.323)
```

Refs. swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5275

Restart the replay from scratch with check_dst activated.
The errors will be stored in another redis db so we could
compare the differences.
As a new consumer group is used, if it's too long to get
to the current replay position, we could start another instance
of the replayers with the previous consumer group.

Related to swh/infra/sysadm-environment#5295

Related to swh/infra/sysadm-environment#4707

Related to swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5275

Related to swh/infra/sysadm-environment#5301