Lazy connectivity for Temporal client and healthCheck

[Spikhalskiy]

Users should be able to create a Temporal Client for a temporarily unavailable Temporal server.

SDKs should have the main client or one of the clients that:

1. Don't have an eager healthCheck and getSystemInfo calls in the constructor.
2. getSystemCapabilities call should be performed on the first client call in a lazy manner.
3. healthCheck endpoint should be exposed on the Client that will allow users to perform a health check for the corresponded gRPC service. It will also allow to initialize a connection in an explicit eager manner and perform initialization activities (system capabilities negotiation right now).
4. There may be a .connect() or .dial call that establishes the connection and fetches system capabilities and which will allow to provide the same functionality/guarantees as an eager client provides.

• Expose health check on WorkflowServiceStubs or WorkflowClient sdk-java#828
• Deprecate NewClient, create Dial, NewLazyClient, and CheckHealth calls and support laziness sdk-go#793
• Expose health check on WorkflowClient sdk-typescript#312
• Lazy connectivity for Temporal client and healthCheck sdk-php#224
• [Feature Request] Lazy connectivity for client and healthCheck sdk-python#92

[bergundy]

@cretz did you expose healthcheck in Go?

[cretz]

@bergundy - Yes, in the linked Go PR from the description. Client.CheckHealth(). I need to open a Core issue to expose it so Python can use it. (EDIT: temporalio/sdk-core#367)

[bergundy]

Thanks @cretz I checked Go off the list