Docs: clarify connection mode for local psql imports in CSV ingestion workflows

I’m running a local CSV normalization + ingest workflow into Supabase Postgres. ### Workflow 1. Parse jobs from local CSV/ZIP sources. 2. Prepare schema and `INSERT ... ON CONFLICT DO NOTHING`. 3. Import via `psql` using Supabase connection details. ### Problem The SQL path is fine, but the workflow gets blocked at connection setup because the docs for import and connection mode are split across different pages. In practice, it’s easy to pick the wrong connection path for local `psql` import unless you already know the connection-mode caveats. ### Expected behavior A single import flow should clearly state: 1. Which connection string/host is recommended for local `psql` import. 2. When that path may fail depending on network/setup. 3. The fallback path to use next. 4. A minimal verified command example. ### Actual behavior I reached a dead-end before SQL execution because the connection details were incomplete/misaligned for my local environment, and I had to infer the correct mode from scattered docs. ### Why this matters This affects practical ingestion pipelines (CSV normalization, dedupe, upsert) where users need to get from local files to database writes quickly. The current docs make that first connection step easier to get wrong than it should be. ### Suggested docs improvement Add a short "Connection mode decision" section in the import guide: 1. If local `psql`, start with X 2. If X fails, use Y 3. Known caveats 4. Minimal end-to-end example