cli: de-duplicate node level information in multi-tenant debug zip

[Internal Discussion](https://cockroachlabs.slack.com/archives/C06301L45NF/p1761155561980739)

Debug zip currently creates a directory for every tenant. It also replicates the structure of a traditional debug zip in that there's a nodes directory for each tenant in the cluster. These node directories can contain duplicative information, which can often be costly from a sizing perspective. Ideally we would de-duplicate the collection of these records, likely by not including them in the `clusters` directory on collection.

Eg: in the below directory, each value of `ranges.json` is the same.
```
├── debug
│   ├── cluster
│   │   ├── bar
│   │   │   ├── nodes
│   │   │   │   └── 1
│   │   │   │   ├── ranges.json
│   │   └── foo
│   │   ├── nodes
│   │   │   └── 1
│   │   │   ├── ranges.json
│   ├── nodes
│   │   └── 1
│   │   ├── ranges.json
```

Jira issue: CRDB-55737

Start by tracing the CLI debug-zip collection path that builds the tenant-specific cluster directories and node records. Compare the example's duplicated ranges.json entries across cluster and top-level nodes. Done means the resulting multi-tenant debug zip avoids redundant node-level information while preserving the needed shared records.

Written by the indexing model from the issue text.