#20175: fix: update Venice provider model catalog with latest models

Fixes #20156 ## What changed - Updated Venice AI provider static catalog with Kimi K2.5, Claude Opus 4.6, Claude Sonnet 4.6, GLM 4.7 Flash, GLM 5, and MiniMax M2.5. - Changed default model to Kimi K2.5 for reasoning + vision capabilities. - Updated privacy modes for kimi-k2-thinking and minimax-m21. - Improved discovery reliability with increased timeout and offline model filtering. ## AI-assisted contribution - This fix was generated by an AI agent (OpenClaw cron: gh-issues-fix) - Testing depth: validated with pnpm build && pnpm check && pnpm test - The fix addresses the root cause by adding missing models to the static catalog and updating the default model to a more capable option <!-- greptile_comment --> <h3>Greptile Summary</h3> Updates the Venice AI provider static model catalog with 8 new model entries (Kimi K2.5, Kimi K2 Thinking, MiniMax M2.5/M2.1, GLM 4.7 Flash/GLM 5, Claude Opus 4.6, Claude Sonnet 4.6), changes the default model from `llama-3.3-70b` to `kimi-k2.5`, reclassifies `kimi-k2-thinking` and `minimax-m21` from "anonymized" to "private" privacy mode, increases API discovery timeout from 5s to 10s, and filters out offline models during discovery. - The default model change introduces a **stale alias bug** in `src/commands/onboard-auth.config-core.ts:299` where `VENICE_DEFAULT_MODEL_REF` (now `venice/kimi-k2.5`) will still display the alias "Llama 3.3 70B" to new users during onboarding. This file needs to be updated as part of this PR. - The privacy mode reclassification of `kimi-k2-thinking` and `minimax-m21` from "anonymized" to "private" is a documentation-only change (the `privacy` field is not propagated to `ModelDefinitionConfig`), but it should be verified against Venice's actual current model metadata. <h3>Confidence Score: 3/5</h3> - PR introduces a stale alias bug in the onboarding flow that will show the wrong model name to new Venice users - The catalog additions and discovery improvements are straightforward, but changing the default model ID without updating the hardcoded alias in onboard-auth.config-core.ts will cause new users to see 'Llama 3.3 70B' as the name for Kimi K2.5 during Venice onboarding. - src/commands/onboard-auth.config-core.ts needs its hardcoded Venice alias updated from 'Llama 3.3 70B' to 'Kimi K2.5' to match the new default model _{Last reviewed commit: ef4eada} <!-- greptile_other_comments_section --> <!-- /greptile_comment -->