feat(models): add GLM-5.2 with 1M context window

[hetdev]

Summary

GLM-5.2 (Zhipu/z.ai) is now available and ships with a truly usable 1M-token context window — a significant upgrade from the 200K window of GLM-5/5.1. This PR registers the model across all relevant surfaces.

Changes

agent/model_metadata.py

• Add "glm-5.2": 1000000 to KNOWN_CONTEXT_LENGTHS. The longest-first substring matching ensures this entry wins over the generic "glm": 202752 catch-all.

hermes_cli/models.py (3 locations)

• OPENROUTER_MODELS: add z-ai/glm-5.2
• _PROVIDER_MODELS curated list: add z-ai/glm-5.2
• _PROVIDER_MODELS["zai"]: add glm-5.2

Source

• GLM-5.2 announcement — Zhipu confirms 1M context window, MIT-licensed open-source release.
• Model architecture: 744B params (40B activated MoE), DeepSeek Sparse Attention (DSA), pre-trained on 28.5T tokens.

Verification

# Substring matching resolves correctly:
# "glm-5.2" → matches "glm-5.2" (len 7) before "glm" (len 3) → 1,000,000 ✅
# "glm-5.1" → matches "glm" only → 202,752 (unchanged) ✅
# "glm-4.5" → matches "glm" only → 202,752 (unchanged) ✅

[teknium1]

Thanks for jumping on GLM-5.2. We merged the broader GLM-5.2 support via #45695, preserving @am423's authorship in git history:

#45695

That landed the 1M context metadata plus the Z.AI model picker/setup/auth probing/fallback surfaces, so this PR is now covered. Closing as duplicate, with appreciation for helping validate the same issue.