BigQuery service released support mutable clustering configuration some time ago, but this hasn't propagated to all clients.

For example, in the docs for python library, there's comments

Clustering fields are immutable after table creation.

Library should ensure:

• Clustering configuration can be modified on a managed table via a table update call
• Documentation reflects this capability, and possibly links to the cloud.google.com content above to describe the behavior when clustering is added/changed.
• Stale comments indicating that clustering requires a partitioning config are removed (you can cluster a table without partitioning it)