Fix missing archived grapher datasets in version tracker

[pabloarosado]

Some archived grapher datasets were missing in VersionTracker.steps_df. The reason was that, when extracting the relevant info from the database, steps were assign either data:// or data-private:// prefixes based on isPrivate. But there were grapher datasets that were manually archived (by setting isArchived and isPrivate to 1) which came from non-private ETL steps.
This PR fixes the issue, and those archived grapher datasets now appear in steps_df.

[owidbot]

Quick links (staging server):

Site	Admin	Wizard

Login: ssh owid@staging-site-fix-missing-datasets-in-version-tracker

chart-diff: ✅

No charts for review.

data-diff: ❌ Found differences

= Dataset garden/health/latest/global_health_mpox
  = Table global_health_mpox
= Dataset garden/who/2024-09-09/flu_test
  = Table flu_test
    ~ Dim country
-       - Removed values: 30 / 71625 (0.04%)
                date     country
          2024-09-30      Bhutan
          2024-09-30   Indonesia
          2024-09-23       Kenya
          2024-09-30    Slovakia
          2024-09-30 Switzerland
    ~ Dim date
-       - Removed values: 30 / 71625 (0.04%)
              country       date
               Bhutan 2024-09-30
            Indonesia 2024-09-30
                Kenya 2024-09-23
             Slovakia 2024-09-30
          Switzerland 2024-09-30
    ~ Column denomcombined (changed data)
-       - Removed values: 30 / 71625 (0.04%)
              country       date  denomcombined
               Bhutan 2024-09-30             55
            Indonesia 2024-09-30             21
                Kenya 2024-09-23             64
             Slovakia 2024-09-30             14
          Switzerland 2024-09-30             37
        ~ Changed values: 29 / 71625 (0.04%)
            country       date  denomcombined -  denomcombined +
              China 2024-09-23            15701            21567
          Indonesia 2024-07-08               52               53
          Indonesia 2024-08-19               70               63
           Malaysia 2024-08-26              930              928
           Malaysia 2024-09-23              924              918
    ~ Column pcnt_poscombined (changed data)
-       - Removed values: 30 / 71625 (0.04%)
              country       date  pcnt_poscombined
               Bhutan 2024-09-30          9.090909
            Indonesia 2024-09-30         28.571428
                Kenya 2024-09-23         21.875000
             Slovakia 2024-09-30         14.285714
          Switzerland 2024-09-30         24.324324
        ~ Changed values: 31 / 71625 (0.04%)
              country       date  pcnt_poscombined -  pcnt_poscombined +
               France 2024-06-17            0.427727            0.452887
            Indonesia 2024-06-24           25.423729           25.000000
            Indonesia 2024-08-12           12.500000           12.195122
             Slovenia 2024-09-16            0.085470            0.085543
          Switzerland 2024-09-16           86.666664           76.666664
= Dataset garden/who/latest/monkeypox
  = Table monkeypox


Legend: +New  ~Modified  -Removed  =Identical  Details
Hint: Run this locally with etl diff REMOTE data/ --include yourdataset --verbose --snippet

Automatically updated datasets matching weekly_wildfires|excess_mortality|covid|fluid|flunet|country_profile|garden/ihme_gbd/2019/gbd_risk are not included

Edited: 2024-10-11 08:20:34 UTC
Execution time: 14.00 seconds

[lucasrodes]

LGTM, thanks Pablo!

I've added some brief docs.

Also a reminder, that we should probably slowly stop using db_conn in favour of OWIDEnv objects. This basically means refactoring some functions in etl.grapher_io and their usage (for the future).