refactor: re-export autoconf config surface from library __init__

Expose the autoconf (Nerves) configuration/serialization surface as a
deliberate public API of each science library, so workspaces and downstream
code can `from autolens import conf` (and jax_wrapper, setup_colab,
setup_notebook, fitsable, with_config, dictable/fitsable/test_mode helpers)
rather than importing the `autoconf` package directly. This keeps the config
layer an implementation detail behind the library namespace.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ciVftxvYpefh59wSkR7jN

Author: claude

autoarray/__init__.py

@@ -107,3 +107,34 @@
 conf.instance.register(__file__)
 
 __version__ = "2026.7.9.1"
+
+# ---------------------------------------------------------------------------
+# Public re-export of the autoconf configuration / serialization surface.
+#
+# Workspaces, tutorials and downstream code import these names from the science
+# library (e.g. ``from autolens import conf``) rather than depending on the
+# ``autoconf`` package directly, so the underlying configuration / serialization
+# layer stays an implementation detail of the library.
+# ---------------------------------------------------------------------------
+from autoconf import conf
+from autoconf import jax_wrapper
+from autoconf import fitsable
+from autoconf import setup_colab
+from autoconf import setup_notebook
+from autoconf.conf import with_config
+from autoconf.dictable import from_dict, from_json, to_dict, output_to_json
+from autoconf.fitsable import (
+    output_to_fits,
+    hdu_list_for_output_from,
+    ndarray_via_fits_from,
+    ndarray_via_hdu_from,
+    header_obj_from,
+)
+from autoconf.test_mode import (
+    with_test_mode_segment,
+    skip_visualization,
+    skip_fit_output,
+    skip_checks,
+    is_test_mode,
+    test_mode_level,
+)