Add data dumps

[ArtOfCode-]

Add data dumps. Weekly export of the entire database minus anything sensitive, uploaded to S3 and made available via a new page linked in the footer. Also adds an option for manually-created dump records, intended for quarterly uploads to Archive.org.

Incorporates #1950 by cherry-pick.

[codecov]

Codecov Report

❌ Patch coverage is 54.60993% with 64 lines in your changes missing coverage. Please review.
✅ Project coverage is 80.30%. Comparing base (2023e74) to head (cfde135).

Files with missing lines	Patch %	Lines
lib/database_changes_checker.rb	0.00%	56 Missing ⚠️
app/jobs/data_dump_job.rb	92.75%	5 Missing ⚠️
app/controllers/dumps_controller.rb	60.00%	2 Missing ⚠️
app/models/dump.rb	87.50%	1 Missing ⚠️

Additional details and impacted files

Components	Coverage Δ
controllers	76.02% <60.00%> (-0.03%)	⬇️
helpers	85.32% <100.00%> (+0.01%)	⬆️
jobs	75.11% <92.95%> (+8.22%)	⬆️
models	93.02% <87.50%> (-0.03%)	⬇️
tasks	61.11% <ø> (ø)

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[github-actions]

Data dump-affecting changes

This pull request changes DB schema and may affect what data is included in the data dump. Please review:

db/migrate/20260120032240_create_dumps.rb

create_table :dumps do |t|

db/migrate/20260513135320_add_defaults_for_data_dumps.rb

change_column_default :filters, :user_id, -1
change_column_default :flags, :escalated, false
change_column_default :users, :sign_in_count, 0
change_column_default :users, :failed_attempts, 0
change_column_default :users, :deleted, false

db/migrate/20260513160917_add_automatic_to_dumps.rb

add_column :dumps, :automatic, :boolean, null: false, default: false

db/migrate/20260513185013_add_link_to_dumps.rb

add_column :dumps, :link, :string

db/migrate/20260519211432_add_checksum_to_dumps.rb

add_column :dumps, :checksum, :string

db/migrate/20260520085943_add_more_default_values.rb

change_column_default :votes, :created_at, '2000-01-01T00:00:00.000000Z'
change_column_default :votes, :updated_at, '2000-01-01T00:00:00.000000Z'

db/migrate/20260520112652_add_filter_default_name.rb

change_column_default :filters, :name, ''

db/migrate/20260520175923_add_user_creation_defaults.rb

change_column_default :users, :created_at, '2000-01-01T00:00:00.000000Z'
change_column_default :users, :updated_at, '2000-01-01T00:00:00.000000Z'
change_column_default :community_users, :created_at, '2000-01-01T00:00:00.000000Z'
change_column_default :community_users, :updated_at, '2000-01-01T00:00:00.000000Z'

[cellio]

I don't recall any details now, but I thought there might have been a case that resulted from timings (maybe a user pressing save on an answer after the question was deleted?), so unless we can rule that out we might get very rare new cases in future.

If that isn't already ruled out, would a potential solution be to have a job to delete such answers, and have it run just before the data dump each week?

Oh, I think you're right. A cleanup job sounds like a good idea (and is separable from this PR).

[cellio]

Tested and works great for me. I read the code changes and understood most but not all of them, so treat this as a functionality review more than a code review.

Edit: Art walked me through the code I was having trouble understanding. LGTM!

[ArtOfCode-]

I have no idea why a sign-in rate limit test has suddenly started failing and have run out of time to work on this for a few days - if anyone else has time to look please feel free.

[cellio]

I have no idea why a sign-in rate limit test has suddenly started failing and have run out of time to work on this for a few days - if anyone else has time to look please feel free.

It's because of the new default creation/update times for users. Don't merge until we resolve.

[cellio]

Retested with the latest changes and all looks good now: new users have the correct (real) timestamps, and data dumps show 1970 for users, community users, and votes. Changing the default in the dump DB only from within the dump job is clever!