Export FLOWS and build from the top-level fluksio package

A test of the declarations had to import them from fluksio.sdk while the
guide teaches `from fluksio import ...`. The worker's inert module gets the
same two names so a module importing them still loads inside a node.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QgG6TJw5qBZzaLghQSRVrq

backend/fluksio/__init__.py

@@ -4,7 +4,8 @@ Two audiences, one import name. In a research repository, ``from fluksio
 import Port, node, Flow`` is the authoring API: it says which of your existing
 functions are nodes and which nodes make up a flow, and ``fluksio sync``
 uploads what it finds. Those names come from :mod:`fluksio.sdk`, which imports
-nothing but the standard library.
+nothing but the standard library. ``FLOWS`` and ``build`` come with them, so a
+test can check the declarations without an engine.
 
 Inside a node, ``import fluksio`` is not this package at all: the worker
 installs a reporter of its own under that name before any node code runs, so
@@ -24,7 +25,7 @@ from importlib.metadata import PackageNotFoundError
 from importlib.metadata import version as _version
 from typing import Any
 
-from fluksio.sdk import Flow, Port, node, use
+from fluksio.sdk import FLOWS, Flow, Port, build, node, use
 
 try:
     __version__ = _version("fluksio")
@@ -32,8 +33,10 @@ except PackageNotFoundError:  # pragma: no cover - a checkout that was never ins
     __version__ = "0.0.0+unknown"
 
 __all__ = [
+    "FLOWS",
     "Flow",
     "Port",
+    "build",
     "emit",
     "load_artifact",
     "node",

backend/tests/sdk/test_build.py

@@ -364,3 +364,15 @@ def test_a_node_in_a_script_run_directly_is_refused_at_sync_not_at_import():
         flow.document()
     with pytest.raises(SyncError, match="run directly"):
         flow.shims()
+
+
+def test_the_registry_and_the_builder_come_with_the_authoring_import():
+    """A test of the declarations needs no second import path."""
+    import fluksio
+    import fluksio.sdk
+
+    flow = a_flow()
+
+    assert fluksio.FLOWS is fluksio.sdk.FLOWS
+    assert fluksio.FLOWS["train"] is flow
+    assert fluksio.build(flow) == fluksio.sdk.build(flow)

docs/getting-started/data-science.md

@@ -415,6 +415,10 @@ finetune = Flow(
 different message, keyword arguments override settings, and `id=` tells two
 uses of one function apart. `train` above is unaffected by any of it.
 
+A test can check the declarations without an engine. Importing the module
+registers its flows in `fluksio.FLOWS`, and `fluksio.build(flow)` returns the
+document `sync` would upload, refusing what the engine would refuse.
+
 ## Sync it
 
 ```sh

worker/fluksio_worker/worker_main.py

@@ -101,6 +101,9 @@ class _Reporter(ModuleType):
     #: The same name the SDK exports at top level, so a node written against
     #: one surface finds it on the other. Set by ``_install_reporter``.
     logger: logging.Logger
+    #: The registry `fluksio sync` reads, empty here: nothing registers into
+    #: it inside a node. Set by ``_install_reporter``.
+    FLOWS: dict[str, Any]
 
     def emit(self, **ports: Any) -> None:
         """Publish on this node's output ports without returning yet.
@@ -174,6 +177,10 @@ class _Reporter(ModuleType):
         """A flow declaration, already read by `fluksio sync`."""
         return _Declared()
 
+    def build(self, *args: Any, **kwargs: Any) -> Any:
+        """A flow document, already built by `fluksio sync`."""
+        return {}
+
 
 class _LogHandler(logging.Handler):
     """Writes to whatever ``sys.stderr`` is at the time of the record.
@@ -409,6 +416,7 @@ def _install_reporter() -> None:
     # its own logging installs a root handler on the same stream.
     logger.propagate = False
     module.logger = logger
+    module.FLOWS = {}
     sys.modules["fluksio"] = module