feat: add schema-compare command to test harness

[markphelps]

Summary

Adds a schema-compare command to the test harness (tools/test-harness/) that compares static (Go tree-sitter) vs runtime (Python) schema generation for regression detection.

How it works

For each model, schema-compare:

1. Builds with COG_STATIC_SCHEMA=1 → extracts OpenAPI schema from Docker label
2. Builds without it → extracts OpenAPI schema from Docker label
3. Compares the two JSON schemas for exact equality
4. Reports a structured diff on mismatch showing the exact JSON paths that diverge

cd tools/test-harness

# Compare all non-GPU models
python -m harness schema-compare --no-gpu --cog-binary /path/to/cog

# Compare a specific fixture
python -m harness schema-compare --model fixture-scalar-types --cog-binary /path/to/cog

Local fixture models

7 local fixture models (fixtures/models/) covering the full input type matrix:

Fixture	Coverage
scalar-types	str, int, float, bool, Secret
optional-types	PEP 604 X | None and Optional[X] for all types
list-types	list[X] and List[X] for str, int, Path, File
optional-list-types	list[X] | None and Optional[List[X]]
constraints-and-choices	ge/le constraints, string/int choices
file-path-types	Path, File, optional Path/File
complex-output	BaseModel structured output

These use repo: local in the manifest and are loaded from fixtures/models/ without cloning.

Example output

Schema Comparison Report (CLI custom)
=====================================

  x fixture-scalar-types      schemas differ
      $.paths./predictions/{prediction_id}/cancel.post.parameters[0].schema.title: value mismatch
        static:  "Prediction_id"
        runtime: "Prediction Id"

No normalization is applied yet — the comparison is strict equality so we can see all differences before deciding which to fix vs which to accept.