Improve ML reactivity when there are many files

[petersg83]

What does it do?

Add indexes for sortable columns of files

Why is it needed?

When files are retrieved, they are sorted by a field like created_at. When there are many files, like 300k, the API response becomes slower (~1 to 2 sec). Adding an index improve the speed of the sort operation.

How to test it?

1. Upload a file to the ML
2. Go to your database, in the files table
3. Copy the file row (without the ID) and paste it 200k times (it's a bit long to do but you can copy several rows at the same time. You can also do an SQL loop but not on SQLite)
4. Go to the ML and check the request time before and after the fix.

Related issue(s)/PR(s)

Fix #14954

[codecov]

Codecov Report

Base: 60.34% // Head: 60.34% // No change to project coverage 👍

Coverage data is based on head (daf3092) compared to base (95ec21c).
Patch has no changes to coverable lines.

Additional details and impacted files

@@           Coverage Diff           @@
##             main   #15221   +/-   ##
=======================================
  Coverage   60.34%   60.34%           
=======================================
  Files        1367     1367           
  Lines       33268    33268           
  Branches     6353     6353           
=======================================
  Hits        20076    20076           
  Misses      11347    11347           
  Partials     1845     1845           

Flag	Coverage Δ
back	50.37% <ø> (ø)
front	64.80% <ø> (ø)
unit_back	50.37% <ø> (ø)
unit_front	64.80% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[Marc-Roig]

Read a conversation somewhere about indexes in strapi.
If I recall correctly, this is not ready for the public yet, but I imagine you can still use this syntax in any of your content-types right?

What work would be needed to make it public? Specify the list of fields you want to index to automatically generate a index name?

[petersg83]

Yes exactly :) and to convert the field name to the database column name (createdAt => updatedAt). Also some thinking to have about columns that don't appear in the user schema (like createdAt and updatedAt)

[Marc-Roig]

Looks good to me, using sqlite I see a very little increment in the disk when the indexes are added indeed.

Seeing this PR I am thinking of preparing some performance tests in the future, to spot which parts are slower and to decide when to use indexes.