feat(controller): optional database migration

[book987]

#4822

Checklist:

• My organization is added to USERS.md.

[CLAassistant]

All committers have signed the CLA.

[alexec]

My thought on this is - how do users know what SQL needs to be applied? Could we modify migrate.go to with a print-only mode - i.e. print out the SQL statements, but not apply them?

[book987]

My thought on this is - how do users know what SQL needs to be applied? Could we modify migrate.go to with a print-only mode - i.e. print out the SQL statements, but not apply them?

🤔 I think print out mode is

1. helpful when using Argo at first time, just get the output and apply somewhere
2. but help little or useless when upgrading Argo, most users don't want to drop and rebuild database i think, they need to find a way generating what they should apply. Even tracking changes from commits since last version is more efficent.
3. also, user should run another kubectl command to get what we print out, and filter SQL from all output by them self

And, like i mention at #4822, i don't think hardcoded db migration in go source code is a good way. I think we can:

• use a seperated directory for db migration like Django do, new schema changes means new migration file, user can get schema chagnes by just seeking at migration directory
• or we can just provide a SQL file represent schema changes when releasing new version

But the ways i mention above is not friendly for user using Argo at first time, Maybe we needs further disign.

[alexec]

I think this is plenty good enough for MVP.