Fix span-metrics' subprocessors bug

[mapno]

What this PR does:
Fixes a bug in which a shared map in the metrics-generator override config was not properly copied to avoid modifying the config of other tenants.

Which issue(s) this PR fixes:
Fixes #

Checklist

• Tests updated
• Documentation added
• CHANGELOG.md updated - the order of entries should be [CHANGE], [FEATURE], [ENHANCEMENT], [BUGFIX]

[joe-elliott]

How difficult is it to construct a test that reproes and then confirm this fixes it?

[mapno]

How difficult is it to construct a test that reproes and then confirm this fixes it?

I think it's feasible. I'll build one 👍

[joe-elliott]

does it make more sense to break the shared memory here:

https://github.com/grafana/tempo/blob/main/modules/generator/config.go#L66

where we copy the config?

[mapno]

does it make more sense to break the shared memory here:

https://github.com/grafana/tempo/blob/main/modules/generator/config.go#L66

where we copy the config?

Yes, but we only need the copy if there are changes to the subprocessors. What you suggests is more consistent, but also creates more allocations 🤷
I'm ok with doing either. If we changed it I don't think the number of allocations would be a major problem.

[joe-elliott]

Yes, but we only need the copy if there are changes to the subprocessors. What you suggests is more consistent, but also creates more allocations 🤷

This is just a few more allocations every couple minutes when it refreshes user overrides? That should be fine. I'd definitely lean towards code clarity over performance in this case.

[mapno]

Changed 👍

[joe-elliott]

looks like a simple test failure, but i'm good on this approach