Java SDK: Pick one of two tasks with a condition

[jason810496]

• depends on: Java SDK: Pack each native Dag's source file into the bundle JAR #74096
• related: Java SDK: Choose one of several tasks with a branch #74439, Java SDK: Trigger another Dag's run from a native Java Dag #74440

Merge order:

1. Java SDK: Pack each native Dag's source file into the bundle JAR #74096 — Pack each native Dag's source file into the bundle JAR
2. Java SDK: Pick one of two tasks with a condition #74438 — Pick one of two tasks with a condition (current one)
3. Java SDK: Choose one of several tasks with a branch #74439 — Choose one of several tasks with a branch
4. Java SDK: Trigger another Dag's run from a native Java Dag #74440 — Trigger another Dag's run from a native Java Dag

Why

A native Java Dag could only run every task it declared. Nothing let a task's result decide what runs next, so a Dag that loads only when there are rows had to be split in two or fall back to a Python stub.

What changes

DagDef.If and @Builder.If declare a task whose boolean names the side that runs; the other is skipped.

// Interface style. With no task id given, one comes from the class: "hasRows".
dag.If(HasRows.class).config("retries", 2).then(load).orElse(reportEmpty);

public class HasRows implements ConditionTask {
  @Override
  public boolean decide(Context context, Client client) {
    return ((Number) client.getXCom("transform")).longValue() > 0;
  }
}

// Annotation style.
@Builder.If(id = "has_rows")
public boolean hasRows(long transformed) {
  return transformed > 0;
}

@Builder.Deps
static class Wiring implements EtlDeps {
  void depends() {
    hasRows(transformed).then(load(transformed)).orElse(loadEmpty());
  }
}

• If is capitalized because if is a Java keyword, and @Builder.If reads the same way.
• The deciding task carries _can_skip_downstream, and the SDK writes the skipmixin_key XCom before the skip, so Airflow skips a cleared side again rather than running it, as Python's SkipMixin does.
• A condition skips only the side not taken, so a task that runs after both sides needs a trigger rule such as none_failed_min_one_success.
• SkipDownstreamTasks is reachable through the transport client only, never through the task-facing Client, so a task body cannot skip tasks the Dag did not declare it may.
• The serialization conformance check now covers the shared branch Dags, and branching is marked supported.

---

Was generative AI tooling used to co-author this PR?

• Yes, with help of Claude Code Opus 5.5 following the guidelines

[jason810496]

Thanks for the review.