feat: configurable thinking truncation via CCBOT_MAX_THINKING_CHARS

[asturwebs]

Summary

• Configurable thinking truncation: Replace hardcoded 500-char limit with CCBOT_MAX_THINKING_CHARS env var. Defaults to 0 (no truncation, full thinking forwarded to Telegram). Set to e.g. 500 to restore compact thinking messages.

Changes

src/ccbot/handlers/response_builder.py

• build_response_parts() gains max_thinking_chars parameter (default 0)
• When max_thinking_chars > 0, truncates thinking content at the configured limit
• When 0 (default), full thinking content is forwarded unchanged

src/ccbot/config.py

• New config field: max_thinking_chars from CCBOT_MAX_THINKING_CHARS env var (default 0)

src/ccbot/bot.py

• Passes config.max_thinking_chars to build_response_parts()

tests/ccbot/handlers/test_response_builder.py

• Tests for default (no truncation), truncation with expandable quotes, and plain text truncation

Usage

# Default: no truncation (full thinking)
ccbot

# Restore old behavior: truncate at 500 chars
CCBOT_MAX_THINKING_CHARS=500 ccbot

Test plan

• All 10 tests pass
• ruff check passes
• ruff format passes (our files)
• pyright — 0 errors

Note: pre-existing ruff format failure on bot.py is unrelated to this PR (see comment).

[asturwebs]

Same as #64 — the 2 failing checks are pre-existing issues on main, not introduced by this PR.

• check (3.12): ruff format --check fails on bot.py — already needs reformatting on the base branch
• check (3.13): Cancelled as a dependency of the above
• claude-review: Internal error in the Claude Code Action, unrelated to this change

All 3 files in this PR (response_builder.py, update.sh, test_response_builder.py) are clean.