Schema artifacts for LLM/tooling contracts (encodings, react settings, url params) via optional dev-only models #1326

Summary: Add first-class JSON Schema artifacts for three public contract surfaces used by LLM/tooling integrations: (1) simple+complex encodings, (2) react settings including apply_encodings keys, (3) url params. Goal is better OSS-discoverable signatures and less downstream drift while keeping runtime behavior unchanged.\n\nConstraints: Do NOT add pydantic as a hard runtime dependency. Keep existing runtime TypedDict + validator flows canonical for execution behavior.\n\nProposal: (1) Add optional dev-only schema models (pydantic v2 behind an extra/dev group). (2) Add schema export script to write committed artifacts: schemas/encodings.schema.json, schemas/react-settings.schema.json, schemas/url-params.schema.json. (3) Add CI drift guard that regenerates and fails on diff. (4) Keep semantic validation in existing validators (JSON Schema is structural; validators keep deeper semantic checks).\n\nSuggested file layout: graphistry/devschemas/encodings.py, graphistry/devschemas/react_settings.py, graphistry/devschemas/url_params.py, graphistry/devschemas/export.py, schemas/*.schema.json committed artifacts.\n\nSource-of-truth mapping today: encodings shape in graphistry/io/types.py; encodings semantics in graphistry/validate/validate_encodings.py; react settings in graphistry/models/surfaces/graphistry_frontend/react_settings.py; url params in graphistry/models/surfaces/graphistry_frontend/url_params.py.\n\nNon-goals: replacing runtime validators with JSON Schema; breaking existing payload compatibility; coupling io metadata wire types directly to GFQL call param types.\n\nAcceptance criteria: optional install path for schema tooling only; committed schemas for the three surfaces above; CI drift check in place; no runtime behavior change for existing APIs/payloads; docs note that schema files are structural contracts while semantic constraints remain in validators.