Polish benchmark scaffold layout

Co-Authored-By: Paperclip <noreply@paperclip.ing>

evals/bench/python/README.md

@@ -0,0 +1,7 @@
+# Python Bridge Layout
+
+This directory reserves the Python side of `paperclip-bench` for ecosystems and evaluator stacks that already expect Python entrypoints.
+
+- `bridge/`: future wrappers around the TypeScript runner and evaluator APIs
+- `evaluators/`: Python-native evaluator adapters when a benchmark ecosystem already ships there
+- `datasets/`: dataset preparation helpers that are easier to keep in Python

evals/bench/python/bridge/.gitkeep

@@ -0,0 +1 @@
+

evals/bench/python/datasets/.gitkeep

@@ -0,0 +1 @@
+

evals/bench/python/evaluators/.gitkeep

@@ -0,0 +1 @@
+

evals/bench/python/src/paperclip_bench/__init__.py

@@ -0,0 +1 @@
+"""Placeholder Python bridge package for paperclip-bench Phase 1."""