Wrap GRPC::CANCELED and DEADLINE_EXCEEDED in an SDK Timeout exception for Update

[drewhoskins-temporal]

Describe the solution you'd like

Make a common exception for update that represents timeout or cancel.

Per-SDK Tickets

• Go - Wrap GRPC::CANCELED and DEADLINE_EXCEEDED in an SDK Timeout exception for Update sdk-go#1479
• Java - Wrap GRPC::CANCELED and DEADLINE_EXCEEDED in an SDK Timeout exception for Update sdk-java#2069
• Core - N/A
• TypeScript - Wrap GRPC::CANCELED and DEADLINE_EXCEEDED in an SDK Timeout exception for Update sdk-typescript#1422
• Python - Wrap GRPC::CANCELED and DEADLINE_EXCEEDED in an SDK Timeout exception for Update sdk-python#529
• .NET - Wrap GRPC::CANCELED and DEADLINE_EXCEEDED in an SDK Timeout exception for Update sdk-dotnet#250
• PHP - Wrap GRPC::CANCELED and DEADLINE_EXCEEDED in an SDK Timeout exception for Update sdk-php#436
• Temporal CLI - N/A

[cretz]

Need to decide what best to call this. Maybe WorkflowUpdateDeadlineExceededX (where X is language specific suffix like Exception or Error)?

[cretz]

It has been decided to do the following:

• Have a new WorkflowUpdateRPCTimeoutOrCanceled exception/error in every language (or similar)
  • If the language supports it, have it inherit a base RPCTimeoutOrCanceled exception/error for future use (or similar)
  • Exception name and docs should make it clear in docs this is for RPC call timeout/cancel and not some concept of "update timeout/cancel" (which may appear someday but does not today)
• Have that thrown/raised when start update or poll for update returns either Canceled or DeadlineExceeded status codes (or if the language has its own cancel exception, catch that)
  • Use the language's wrapping mechanism to wrap the gRPC exception/error instead of replace it
• Confirm via tests that a timed out update uses this exception/error by using a low timeout in an update that doesn't complete for a long time