v0.2.0: Brand Bibel und Multi-Provider ueber alle Tier-1-Templates

Added

• internal style guide as the single source of truth for tone, structure, branding, and quality gates across every template in this repo.
• Multi-provider switch in template 01 (Voice Agent Cross-Session Memory). A Set Provider + Route by Provider pair lets builders choose OpenAI (default) or Anthropic, or extend with a third branch (Gemini, Mistral, local Ollama). Both branches converge in Normalize LLM Output so memory writes stay identical regardless of provider.
• .github/ standard scaffolding: FUNDING.yml, ISSUE_TEMPLATE (bug + template-request + config), PULL_REQUEST_TEMPLATE, and a validate-workflows GitHub Action that catches missing refs, real credentials, and em-dashes before merge.
• templates/_TEMPLATE/ skeleton for new templates so every contribution starts at the internal style guide baseline.
• CHANGELOG.md, CODE_OF_CONDUCT.md, SECURITY.md, ECOSYSTEM.md as standard repo top-level files.

Changed

• Default OpenAI model in template 01 from gpt-4o-mini (deprecated) to gpt-5-mini. Anthropic stays on claude-haiku-4-5 for the speed-tier branch.
• Template 01 README rewritten to reflect the new multi-provider pattern, with a clearer architecture diagram and explicit instructions for adding additional LLM branches.

Fixed

• Validation guard in CI rejects PRs where workflow.json has missing node references, embedded test pinData, or em-dashes in markdown / workflow files.