Initial import: XAUUSD news sentiment models, evals, and data tooling

TF-IDF baseline (news_sentiment.py, models/xauusd_sentiment.joblib) and the
two-head DeBERTa-v3-small transformer (transformer/, news_transformer.py)
with its Google News crawler and DeepSeek labeller (data_collect/), the
golden-set and archive evals (eval/), and the 2026-09-22 and 2026-09-27
results and progress notes.

Transformer weights (model.pt, model.onnx) and the crawl database are not
tracked; rebuild with transformer/train.py and transformer/export_onnx.py.
The DeepSeek label cache in data/ is tracked.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

requirements-train.txt

@@ -0,0 +1,13 @@
+# GPU training environment (.venv-train). The CPU runtime needs only
+# requirements.txt; nothing here is needed to score headlines.
+torch
+transformers
+sentencepiece
+tiktoken
+protobuf
+onnx
+onnxruntime
+onnxscript
+scikit-learn
+numpy
+pandas  # eval scripts (spearman)