fix(providers): use warehouse-backed storage

[jonathanhaaswriter]

Summary

• replace provider storage wiring from direct Snowflake clients to the generic warehouse interface
• update provider table sync helpers to use dialect-aware warehouse merge/upsert behavior
• add regression coverage for provider syncs against a non-Snowflake warehouse backend

Validation

• go test ./internal/providers -run "TestBaseProviderSyncTable_UsesConfiguredWarehouse|TestEnsureProviderTable_UsesIdempotentAlter|TestEnsureProviderTable_SkipsExistingColumnsCaseInsensitive"
• go test ./internal/app ./internal/providers
• go test -race ./...
• GOTOOLCHAIN=go1.26.1 go run github.com/golangci/golangci-lint/v2/cmd/golangci-lint@v2.8.0 run --timeout 5m ./...

[jonathanhaaswriter]

One thing I think we missed here: the provider cleanup delete path in internal/providers/storage.go still hard-codes ? placeholders (DELETE ... IN (?, ?, ...)) after switching to warehouse-backed storage. Inserts/upserts were moved to dialect-aware helpers, but this delete path wasn't, so Postgres deletes look like they'll fail and leave stale provider rows behind.

Can we build that SQL with the warehouse placeholder helper (or move it into a dialect-aware tableops helper) before landing?