Pin text I/O to UTF-8 and fail CI on locale-dependent reads/writes - #3296

Stacked on #3245 (its commit is included here; this PR shrinks to one commit once that merges). Review the second commit only.

Motivation and Context

#3244 / #3245 fix the two read_text() calls the test suite trips over on a Windows machine whose ANSI code page isn't cp1252. The underlying problem is wider than those two lines: any text-mode open() / Path.read_text() / write_text() / tempfile / subprocess(text=True) without encoding= uses the process locale, and CI structurally can't notice — ubuntu is UTF-8, and windows-latest is cp1252, which happens to decode every non-ASCII byte we currently have (em dashes, box drawing, arrows) as mojibake rather than raising. On cp936/cp932/cp949/cp950 the same bytes are illegal sequences.

The one place this reaches users is mcp install: src/mcp/cli/claude.py read claude_desktop_config.json with the locale codec and wrote it back through json.dumps, so on a stock cp1252 or a CJK Windows box, pre-existing entries containing non-ASCII (a C:\Users\书清\… path, a server named météo) were either rejected or silently persisted as mojibake while the command reported success. The stories harness (examples/stories/_harness.py) also reads the same manifest #3245 fixes in the tests, so every python -m stories.<name>.client README command failed on such a machine, as did the two repo scripts.

This PR:

• passes encoding="utf-8" at every remaining call site (cli, stories harness, scripts, tests, simple-chatbot, one docs_src snippet); the two diagnostic subprocess.run(..., text=True) captures become encoding="utf-8", errors="replace"; and drops a ✓ from update_readme_snippets.py's output, which raised UnicodeEncodeError whenever stdout was a non-UTF-8 pipe (e.g. under pre-commit on Windows).
• makes the class fail CI from now on: the pytest step and scripts/test set PYTHONWARNDEFAULTENCODING=1 (PEP 597), so an omitted encoding= raises EncodingWarning at the call site on every OS, which our existing filterwarnings = ["error"] makes fatal (env var rather than -X so pytest-xdist workers inherit it; one scoped ignore for pytest_examples' own Popen). The two script steps run with -X warn_default_encoding -W error::EncodingWarning. ruff's PLW1514 is enabled as the edit-time counterpart (preview rule, hence preview = true + explicit-preview-rules = true so nothing else changes); on its own it only recognises obvious Path(...) receivers and misses e.g. (a / "b").read_text(), which is why the runtime guard is the primary mechanism.
• upgrades test_other_servers_preserved to seed non-ASCII entries, so the mcp install round-trip is covered directly.

How Has This Been Tested?

• Unmodified main with only the env var set fails deterministically at tests/examples/conftest.py:41 (EncodingWarning: 'encoding' argument not specified) on Linux and on windows-latest; with this branch ./scripts/test is green (5582 passed, 100% coverage, strict-no-cover clean).
• Full suite with this change under PYTHONWARNDEFAULTENCODING=1 on windows-latest (3.10 and 3.12), both at the stock code page 1252 and with the runner switched to 936: green, no additional warning sources.
• Under generated zh_CN.GBK / en_US.CP1252 glibc locales (PYTHONUTF8=0): drove mcp install against a config seeded with Chinese and accented entries — before: cp1252 rewrites them as 文件系统 / José, GBK fails with 'gbk' codec can't decode byte 0x94; after: entries byte-identical, new server added, re-install merges env vars. python -m stories.tools.client and both scripts' --check go from UnicodeDecodeError to OK under GBK.
• The upgraded test fails on the old claude.py under cp1252, under GBK, and under the warning guard on a UTF-8 box.

Breaking Changes

None. mcp install writes the same bytes as before (json.dumps still ASCII-escapes); it just decodes the existing file correctly first.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update

Checklist

• I have read the MCP Documentation
• My code follows the repository's style guidelines
• New and existing tests pass locally
• I have added appropriate error handling
• I have added or updated documentation as needed

Additional context

• Python 3.15 makes UTF-8 mode the default (PEP 686), which retires the failure mode for users on 3.15+, but 3.10–3.14 stay supported for a long while and the explicit argument remains correct after that.
• Not done here: running one Windows matrix cell under code page 936 (Set-WinSystemLocale zh-CN takes effect for new processes on the hosted image without a reboot). It would cover third-party code and spawned children, which the warning can't reach; happy to add it if we think it earns its keep.

_{AI Disclaimer}