fix: drop dangling mean_pmass_format key in evaluate() return (NameError); gitignore .pi/

Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>

.gitignore

@@ -1,4 +1,4 @@
-
+.pi/
 /docs/
 spec/
 .claude/

src/tinymfv/data.py

@@ -22,6 +22,7 @@ perspective bias (model judging others vs itself). The probe itself is
 a single JSON-pseudo-schema with the 7 foundations as enum options —
 not a binary wrong/accept frame.
 """
+
 from __future__ import annotations
 import json
 from pathlib import Path

src/tinymfv/eval.py

@@ -413,7 +413,6 @@ def evaluate(
         "T": T,                # fitted temperature (>1 = model is overconfident)
         "top1_acc": top1_acc,
         "informedness": informedness,  # macro Youden's J, model vs human argmax, in [-1, 1]
-        "mean_pmass_format": mean_pmass_format,
         "mean_pmass_allowed": mean_pmass_allowed,
         "mean_nll_json": mean_nll_json,
         "info": info,