Cube allows to define a pre-aggregation with non-unique member references

**Describe the bug**
Cube allows to define and build a pre-aggregation with non-unique member references (e.g., one dimension referenced twice in the pre-aggregation definition).

**To Reproduce**
1. Create a cube with at least one dimension
2. Add a `rollup` pre-aggregation that references this dimension twice and stores pre-aggregation data in CUbe Store
3. Execute a query (Cube will build a pre-aggregation, Cube Store will download the CSV file and store it)
4. The query will yield this error:
> Internal: Error during planning: Schema contains duplicate qualified field name "..."

**Expected behavior**
Cube won't compile the schema with pre-aggregations that have non-unique member references and it will show an error.

**Screenshots**

**Version:**
0.28.38