fix: smart truncation + --verbose for --list output (#35)

* fix: smart truncation + --verbose + --summarize for --list output

Fixes #32

- Default: word-boundary truncation with '...' instead of hard character slice
- --verbose: full descriptions wrapped with textwrap, continuation lines aligned
- --summarize: AI one-sentence summaries via OPENAI_API_KEY or ANTHROPIC_API_KEY
  Model override: MCP2CLI_SUMMARIZE_MODEL env var (no new dependencies)
- Applied consistently to list_mcp_commands, list_openapi_commands,
  list_graphql_commands and all their call paths

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* Remove --summarize feature, keep smart truncation + --verbose

The --summarize feature (LLM-generated summaries) has scaling issues
(no batching, token limits) and fragile JSON parsing of LLM output.
Remove it for now; smart word-boundary truncation and --verbose wrapping
are solid and ready to ship.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* Bump version to 2.5.0

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Parsa Jorat <nokashiarika@gmail.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>