docs: Comprehensive Spice.ai connector docs (Cloud + Spice-to-Spice / sidecar)

[claudespice]

Summary

Rewrite the Spice.ai data connector docs to cover both architectures the connector actually supports — the existing docs only covered the managed Cloud Platform path. The same connector federates queries to any Spice runtime over Arrow Flight, including self-hosted Spice instances in a cluster-sidecar, edge-to-core, or local-dev topology.

What changed

website/docs/components/data-connectors/spiceai/index.md (rewritten)

• Both topologies up front: Spice → Spice Cloud Platform vs. Spice → Spice (self-hosted / sidecar), with quick-start examples for each.
• All five from: URI formats: path-style, colon-style, and URL-style cloud forms; plus colon-prefixed http:// / https:// / grpc+tls:// forms for self-hosted upstreams.
• Full parameter table with the actual user-facing names: spiceai_api_key, spiceai_token, spiceai_region, spiceai_endpoint, spiceai_flight_endpoint, spiceai_tls_ca_certificate_file. The previous docs incorrectly used unprefixed names like endpoint and were missing spiceai_region and spiceai_tls_ca_certificate_file entirely.
• Endpoint resolution precedence documented: explicit spiceai_endpoint → from: URI scheme → built from spiceai_region.
• Spice → Spice authentication via the upstream's runtime.auth.api-key.keys (not previously documented).
• TLS with private CAs via spiceai_tls_ca_certificate_file for cluster-internal certs.
• Append-stream CDC between Spice runtimes with refresh_mode: append example.
• Worked sidecar example including TLS, CA pinning, append streams, and a multi-dataset spicepod.

website/docs/components/data-connectors/spiceai/deployment.md (rewritten)

• Topology decision matrix mapping each use case to its endpoint format.
• Authentication source table distinguishing Cloud (always required) from self-hosted (optional, must match upstream runtime.auth.api-key.keys).
• Sidecar deployment pattern for Kubernetes — co-locating a sidecar Spice with each application pod and federating to a central upstream.
• Edge → core federation pattern where edge runtimes accelerate local datasets and federate cold queries to a core Spice.
• TLS / CA pinning operations guidance.
• Per-topology troubleshooting: new error rows for UnsupportedEndpointScheme, CloudEndpointRegionMismatch, MissingRequiredParameter: api_key or token, TLS handshake failures with private CAs.

Verification

All claims verified against the actual connector source:

• from: URI parsing → spice_dataset_path (crates/runtime/src/dataconnector/spiceai.rs:526-553); test cases at :651-757.
• Self-hosted endpoint detection → is_flight_endpoint_path (:207-212).
• Supported schemes (http://, https://, grpc+tls://; rejects grpc://) → ensure_supported_endpoint_scheme (:214-223).
• Parameter list → PARAMETERS array (:176-184).
• Endpoint resolution order → get_endpoint (:252-279).
• Authentication fallback to anonymous on self-hosted → get_credentials (:293-306); test (:858-864).
• Real-world parameter names confirmed in test/spicepods/tpch/sf1/federated/spicecloud.yaml.

Test plan

• Verified parameter names and defaults against Rust source
• Verified URI-format parsing against connector tests
• Confirmed runtime.auth.api-key.keys syntax matches runtime.md reference
• Cross-links resolve
• Markdown renders correctly

[github-actions]

🔍 Pull with Spice Failed

Passing checks:

• ✅ Title meets minimum length requirement (10 characters)
• ✅ No banned labels detected

Failed checks:

• ❌ Please select at least one of the required labels for this pull request: area/blog, area/docs, area/cookbook, dependencies
• ❌ At least one assignee is required for this pull request.

Please address these issues and update your pull request.

[github-actions]

🚀 deployed to https://5261d810.spiceai-org-website.pages.dev