How should a data engineer configure a DLT pipeline with the following streaming tables to retain manually deleted or updated records in the raw_iot table (which ingests raw device measurements from a heart rate tracker) while ensuring the downstream bpm_stats table (which computes user statistics from BPM measurements) is recomputed during pipeline updates?

The pipeline consists of:

• raw_iot: Ingests raw heart rate device measurements
• bpm_stats: Incrementally computes user BPM statistics from raw_iot