BigQuery SQL : Decrease remaining minutes counter in successive rows calculation

Question

I have some logs with dates and an aggregated minutes counter time.

Each hours (a row) had been filled out with a trick like this one : Duplicate groups