feat: update DEFAULT_AI_MODEL default in agentic.mk from gpt-4.1 to claude-sonnet-4.5

[tschm]

Summary

The DEFAULT_AI_MODEL variable in agentic.mk currently defaults to gpt-4.1. jquantstats overrides this to claude-sonnet-4.5 in its repo-owned Makefile. Given that analyse-repo and install-claude already target Claude Code CLI as the primary agentic tool (see #899), the default model should reflect this.

Current state in agentic.mk

DEFAULT_AI_MODEL ?= gpt-4.1

Proposed change

DEFAULT_AI_MODEL ?= claude-sonnet-4.5

Rationale

• The analyse-repo target uses the Claude CLI, not Copilot — so the default model should be a Claude model.
• claude-sonnet-4.5 is the current recommended Claude model for agentic code tasks.
• Projects that prefer a different model (e.g. GPT-based via Copilot) can still override this in their own Makefile.

Related

• fix: analyse-repo target should use Claude CLI instead of Copilot #899 — fix: analyse-repo target should use Claude CLI instead of Copilot