429 Too Many Requests on bulk user updates — how to implement proper backoff

I can’t seem to figure out why the PUT /api/v2/users/{userId} endpoint returns HTTP 429 when executing a parallel Terraform plan for bulk user modifications, despite adhering to documented rate limits.

The error payload indicates:

{"error":"too_many_requests","message":"Rate limit exceeded"}

What is the correct backoff strategy to implement in the provider wrapper to prevent this failure during large-scale deployments?