TaskQueue rebuilds should cancel GCB before timeout

Google Cloud Tasks with an HTTP target have a timeout of up to 30 minutes: https://cloud.google.com/tasks/docs/creating-http-target-tasks#handler

Most rebuilds complete within this timeframe, but for the ones that don't complete in this time we should release the resources before the timeout happens. In particular we should set a context with timeout at the beginning of the API handler (only for task-queue requests) and that context should be responsible for cancelling the GCB invokations.

I would also be interested to understand better how the cloud run invokcation handles the task queue cancelling its request. Does cloud run terminate immediately? Run until it hits its own timeout?

Start at the API handler and the path that identifies task-queue requests. Trace how Google Cloud Build invocations are cancelled and how Cloud Run handles a cancelled task request; done means a task-queue rebuild releases its resources before the Cloud Tasks timeout, with the cancellation behavior understood.

Written by the indexing model from the issue text.