[Copilot]

Issue describing the changes in this PR

Resolves #311.

Support restarting an orchestration from the beginning with its original name and input, a required target instance ID, and an optional version override. Execution history is not copied.

Pull request checklist

• My changes do not require documentation changes
  • Otherwise: Documentation issue linked to PR
• My changes are added to the CHANGELOG.md
• I have added all required tests (Unit tests, E2E tests)

Additional information

• Add restartInstance(instanceId, RestartInstanceOptions) with fluent setNewInstanceId and setVersion options. Null options and missing or blank target IDs are rejected before the RPC.
• Preserve the existing restartInstance(instanceId, boolean) overload and its behavior.
• An omitted or null version preserves the backend's default restart behavior; "" explicitly selects an unversioned orchestration.
• Add nullable TaskOrchestrationContext.getSourceInstanceId(). Clones expose their immediate source during execution and replay; sub-orchestrations do not inherit this reference.
• Use the backend restart RPC without falling back to scheduling a new orchestration. Requires a backend supporting explicit restart targets and versions.
• Pin protobuf generation to the upstream explicit-target/version contract.

client.restartInstance("order-123",
    new RestartInstanceOptions()
        .setNewInstanceId("order-123-retry")
        .setVersion("2.0"));

// Inside the clone:
ctx.getInstanceId();       // "order-123-retry"
ctx.getSourceInstanceId(); // "order-123"
ctx.getVersion();          // "2.0"

Cloning order-123-retry again exposes order-123-retry as the immediate source, rather than order-123.

Focused tests cover version presence, required options and IDs, chained clones, backend errors, legacy behavior, and replay/sub-orchestration isolation.