Fix Windows encoding failure during project generation

[markbackman]

Summary

• Path.write_text / read_text in src/pipecat_cli/generators/project.py relied on the platform default encoding. On Windows that is cp1252, which cannot encode the → characters in cascade-mode templates (templates/server/bot_cascade.py.jinja2:9 and templates/README.md.jinja2:3), so pipecat init quickstart failed mid-generation with 'charmap' codec can't encode character '→' and left a partial project directory.
• Added encoding="utf-8" to all 11 text I/O sites in generators/project.py (10 writes + 1 template read). No new abstraction — UTF-8 is named inline at each call site.
• Added a regression test test_generation_uses_utf8_on_windows_locale that monkeypatches Path.write_text / read_text to simulate the Windows cp1252 fallback (data.encode("cp1252") raises on →), then runs the quickstart configuration end-to-end and asserts both server/bot.py and README.md were written with the arrow intact.
• Added an ## [Unreleased] entry in CHANGELOG.md.

Fixes pipecat-ai/pipecat#4523.

Test plan

• uv run pytest tests/test_project_generation.py tests/test_quickstart.py tests/test_client_generation.py — 49 passed.
• New regression test passes with the fix and fails (UnicodeEncodeError: 'charmap' codec can't encode character '→' in position 182) when generators/project.py is reverted — same failure mode as the issue report.
• uv run ruff check and uv run ruff format --check clean on touched files.
• Manual confirmation on a real Windows host (no Windows VM available locally; the monkeypatch reproduces the failure faithfully on macOS).

🤖 Generated with Claude Code