Fix flaky data-pull tests by truncating timestamps

[zachmargolis]

🛠 Summary of changes

Follow-up to #8359 (comment)

Example spec failure, this works around it by truncating timestamps to have less drift (usually this comes from roundtrips to the database)

       (compared using ==)
       Diff:
       @@ -1,9 +1,9 @@
        [["uuid", "email", "confirmed_at"],
         ["16ccac88-7993-4cbe-9cc1-06ca84948720",
          "user246@example.com",
       -  2023-05-09 12:53:00.664757000 -0600],
       +  2023-05-09 12:53:00.664756000 -0600],

[mitchellhenke]

Thank you!

[zachmargolis]

Thank you!

Ya no worries. I thought my original .round(6) would have fixed it but I guess... not! We can always round even more if needed or switch to like kind_of(Time) matchers in the future

[zachmargolis]

Thank you!

Ya no worries. I thought my original .round(6) would have fixed it but I guess... not! We can always round even more if needed or switch to like kind_of(Time) matchers in the future

ended up going for that approach in c3c10d4