[Feature flag] Enable :project_export_as_ndjson feature flag on staging and measure results

What

Since we have merged our NDJSON work on master (&2734)

We should enable project_export_as_ndjson feature flag on staging, for 'export-load-test' and 'import_export_test' namespace and run project export rake task in order to test, and measure its performance.

We should test export with and without feature flag, and compare memory consumption when exporting big (export-load-test/glhq-test-import-oo) and small project (import_export_test/small_project).

Roll Out Steps

1. Run export rake task while project_export_as_ndjson is disabled

sudo gitlab-rake 'gitlab:import_export:export[user_name, export-load-test, glhq-test-import-00, /path/to/legacy_export_big.tar.gz, true]'

2. Run export rake task for the smaller project

sudo gitlab-rake 'gitlab:import_export:export[user_name, import_export_test, small_project, /path/to/legacy_export_small.tar.gz, true]'

3. Enable project_export_as_ndjson feature flag for export-load-test/glhq-test-import-00 project
4. Run again export rake task

sudo gitlab-rake 'gitlab:import_export:export[user_name, export-load-test, glhq-test-import-00, /path/to/ndjson_export_big.tar.gz, true]'

5. Enable project_export_as_ndjson feature flag for import_export_test/small_project project
6. Run export rake task for the smaller project

sudo gitlab-rake 'gitlab:import_export:export[user_name, import_export_test, small_project, /path/to/ndjson_export_small.tar.gz, true]'

7. Test exporting import_export_testsmall_project/ from UI
8. Compare measurement results from 1. 4. and 6. and add them as a comment to this issue
9. Disable feature flag on staging

Owners

• Team: Memory team
• Most appropriate slack channel to reach out to: #g_memory
• Best individual to reach out to: @nmilojevic1

Expectations

We don't expect a big improvement in execution time during export, but we should have constant memory consumption.