Enable compaction and retention in Tanka single-binary

[irizzant]

Describe the bug
As far as I can see retention is enabled in microservices mode but in single-binary only compacted_block_retention is provided.

To Reproduce
Steps to reproduce the behavior:

1. Start Tempo in single-binary mode with Tanka
2. Inspect the ConfigMap

Expected behavior
Retention enabled by default

Environment:

• Infrastructure: Kubernetes
• Deployment tool: Tanka

[joe-elliott]

That line in the single binary jsonnet is really strange. I'm not sure why we are setting it to 24h. My guess is it should be removed and allow the default of 1h to be used. I'm wondering if whoever made it confused it with block_retention when they created it (probably me).

compacted_block_retention is configurable value that controls how long compacted blocks are left in object storage after they have been compacted into larger blocks. They are generally left in object storage for an hour or so. This allows all the different polling components (queriers and compactors) to update their state to see the new blocks before the old blocks are removed.

I think we should just change compacted_block_retention -> block_retention. Thoughts?