SDK should fail workflow task if payloads size it known to be too large

Is your feature request related to a problem? Please describe.

Currently Temporal has limits on the max payload size that can be used for operations like starting a workflow or signaling a workflow. If a user hits this limit the request will fail, leaving the workflow run in a failed state. If a workflow can be updated to minimize the payload size (e.g. compress data, move to some other storage), existing workflow runs can be resumed.

Describe the solution you'd like

If the SDK knows the error limits for payloads, it can submit a workflow task failure rather than submitting a completion that contains oversized payloads. This saves network costs by avoiding the upload of known oversized payloads and allows workflows to continue, provided the workflow is updated to address the oversized payloads.

Additional context

Per-SDK Tickets

• temporalio/sdk-java#2833
• temporalio/sdk-core#1201
• temporalio/sdk-dotnet#638
• temporalio/sdk-ruby#413
• temporalio/sdk-php#801
• Temporal CLI - N/A

Start with the Temporal payload-size limits linked in the issue and review the per-SDK tickets for sdk-core, sdk-dotnet, and sdk-ruby, along with the still-open Java and PHP tickets. Identify where each SDK decides between workflow-task completion and failure; done means known oversized payloads are not uploaded in a completion and the task is failed instead.