Support CancellationTokens for cancelling running operations

Being a public SDK, all externally visible interfaces should support operation cancellation (unless it will send the system into a broken state). We have at times seen it take a long time for blob upload causing us at times to cancel it (because we aren't sure at that point if it is simply stuck or just being insanely slow). Retrying to queue an orchestration at this point at times results into having multiple orchestrations . This would at times be fine, but if the orchestration has been picked up for processing, by the time orchestration is re-queued, operations might already have been executed.

This is just one example. There are other examples like shutting down of service etc where various operations might need to be cancellled.

Inizia facendo un inventario delle interfacce pubbliche dell’SDK coinvolte nel caricamento dei blob, nell’orchestrazione delle code e nell’arresto del servizio. Determina quali operazioni possono accettare CancellationTokens in sicurezza e definisci cosa significhi il completamento per ciascuna; l’issue è completata quando le operazioni visibili dall’esterno supportano la cancellazione, a meno che la cancellazione possa lasciare il sistema non funzionante.

Scritto dal modello di indicizzazione a partire dal testo della issue.