Define Catalyst generic SQL architecture and Spark acceptance

[pmanko]

Summary

• Define Catalyst as a generic SQL-connected application with an explicit dialect and the complete schema readable through the configured connection.
• Keep validation advisory and require the exact selected SQL to return rows or the database error.
• Define FHIR Data Pipes to Parquet to Spark SQL as the selected reference deployment, separate from Catalyst core.
• Require one visible refused write attempt as acceptance evidence for the Spark connection.
• Remove obsolete deployment guidance, duplicate Dashboard contracts, and the stale approved-view golden-query fixture.

This pull request aligns documentation and contracts. Generic-connection and Spark runtime implementation remain open work.

Validation

• Catalyst Gateway: 351 tests passed.
• Catalyst UI: 242 tests passed; type checking and production build passed.
• Catalyst analytics contracts: 51 tests passed, 1 environment-dependent test skipped.
• Diff checks passed.

The dependent harness pull request must merge after this one so its Catalyst revision is already on main.