Add glm-5.3 model family entry

[jatmn]

Summary

Adds an exact model-family entry for Z.ai's GLM-5.3 model, which was previously only matched by the broad glm-5* pattern.

Changes

• configs/model-families/z-ai.toml: Added z_ai_glm_5_3 entry with:
  • 1M context window (same as GLM-5.2)
  • Text-only input modality
  • auto_review_model_override = "glm-5.3"
  • Proper reasoning_effort forwarding alongside thinking.type (like GLM-5.2)
• src/config_tests.rs: Added assertions that the entry loads and applies the correct reasoning_effort transform
• README.md and docs/model-family-catalogs.md: Updated model coverage table

Why needed

Previously, glm-5.3 was only matched by the broad glm-5* pattern which:

• Used the wrong context window (200k from glm-5/5.1 base)
• Did not forward reasoning_effort natively
• Had incorrect auto_review_model_override

This caused issues with network escalation requests being auto-denied due to incorrect or missing model metadata.

Validation

• cargo test --locked (647 passed, 0 failed)
• cargo build --locked
• cargo clippy --locked --all-targets --all-features -- -D warnings (clean)
• git diff --check (clean)

Summary by Sourcery

Add first-class GLM-5.3 support with accurate metadata and provider-compatible reasoning transformations.

New Features:

• Add an exact GLM-5.3 model-family configuration with its metadata, reasoning capabilities, and provider override.

Bug Fixes:

• Correct GLM-5.3 context, model override, and reasoning behavior that was previously inherited from the broader GLM-5 family configuration.

Enhancements:

• Support reasoning-effort aliases and native Responses request normalization for provider-specific effort levels.
• Add typed static boolean request morphs with configuration validation.

Documentation:

• Document GLM-5.3 model coverage and its reasoning transformation behavior.
• Document reasoning-effort aliases and static boolean morphs.

Tests:

• Add coverage for GLM-5.3 metadata, reasoning transformations, native requests, and tool-continuation history.
• Add validation and runtime tests for typed static morph values and reasoning-effort aliases.

[sourcery-ai]

Hey - I've left some high level feedback:

• The new GLM-5.3 reasoning_effort assertions in first_class_reasoning_and_tool_translation_for_target_models are nearly identical to the GLM-5.2 ones; consider extracting a small helper that validates a family’s append_chat_request_morphs to avoid duplication and keep future additions cheaper.

Prompt for AI Agents

Please address the comments from this code review:

## Overall Comments
- The new GLM-5.3 reasoning_effort assertions in `first_class_reasoning_and_tool_translation_for_target_models` are nearly identical to the GLM-5.2 ones; consider extracting a small helper that validates a family’s `append_chat_request_morphs` to avoid duplication and keep future additions cheaper.

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}

[sourcery-ai]

Hey - I've left some high level feedback:

• In assert_has_append_morph, the panic message "{label} family exists" reads like a successful condition; consider changing it to something like "{label} family should exist" or using expect for a clearer failure message.
• The helper name assert_has_append_morph is a bit generic given it specifically checks for the reasoning.effort -> reasoning_effort rename; renaming it to something like assert_has_reasoning_effort_append_morph would make the intent clearer at the call sites.

Prompt for AI Agents

Please address the comments from this code review:

## Overall Comments
- In `assert_has_append_morph`, the panic message `"{label} family exists"` reads like a successful condition; consider changing it to something like `"{label} family should exist"` or using `expect` for a clearer failure message.
- The helper name `assert_has_append_morph` is a bit generic given it specifically checks for the `reasoning.effort -> reasoning_effort` rename; renaming it to something like `assert_has_reasoning_effort_append_morph` would make the intent clearer at the call sites.

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}