fix(continuous-learning-v2): reconfigure stdout to UTF-8 on Windows

`instinct-cli.py status` prints a confidence bar using the box-drawing
characters █ and ░. On Windows the default console codec is cp1250 /
cp1252, and Python's stdout inherits it, so the print statement crashes
with:

    UnicodeEncodeError: 'charmap' codec can't encode characters in
    position 4-13: character maps to <undefined>

This affects any Windows user whose console is not manually set to
`chcp 65001`, even after #353 (file I/O UTF-8) and #216 (.md glob) —
those fixes cover reading instinct files, but the crash happens on the
*output* side when rendering the status table.

Fix: reconfigure `sys.stdout` / `sys.stderr` to UTF-8 at startup, guarded
by `sys.platform == "win32"`. The reconfigure call is wrapped in
`try/except` because a caller may pipe into something that doesn't
support reconfigure (e.g., a captured buffer in tests).

No-op on Linux/macOS.

Verified locally on Windows 11 / Python 3.14 / cp1250 console:
`instinct-cli.py status` now prints all 155 instincts from
`~/.claude/homunculus/instincts/` without crashing.

Author: georgeradcj

skills/continuous-learning-v2/scripts/instinct-cli.py

@@ -28,6 +28,16 @@
 from collections import defaultdict
 from typing import Optional
 
+# Force UTF-8 stdout/stderr on Windows so the confidence bar (█░) and any
+# non-ASCII instinct metadata don't crash `status` under cp1250/cp1252.
+# No-op on Linux/macOS, where stdout is already UTF-8 by default.
+if sys.platform == "win32":
+    try:
+        sys.stdout.reconfigure(encoding="utf-8")
+        sys.stderr.reconfigure(encoding="utf-8")
+    except Exception:
+        pass
+
 try:
     import fcntl
     _HAS_FCNTL = True