fix: skip Docker restart for Dremio to preserve Nessie metadata cache

After docker compose stop/start for seed caching, Dremio loses its
in-memory metadata cache. The VDS view validator then cannot resolve
Nessie-backed tables, causing all integration model views to fail.

Since the Dremio external seeder with COPY INTO is already fast (~1 min),
seed caching provides no meaningful benefit. Excluding Dremio from the
Docker restart eliminates the metadata cache loss entirely.

Co-Authored-By: Itamar Hartstein <haritamar@gmail.com>

Author: devin-ai-integration[bot]

.github/workflows/test-warehouse.yml

@@ -317,7 +317,7 @@ jobs:
         run: dbt seed -f --target "${{ inputs.warehouse-type }}"
 
       - name: Save seed cache from Docker volumes
-        if: steps.seed-cache.outputs.cache-hit != 'true' && inputs.warehouse-type != 'duckdb' && inputs.warehouse-type != 'trino' && inputs.warehouse-type != 'spark' && (inputs.warehouse-type == 'postgres' || inputs.warehouse-type == 'clickhouse' || inputs.warehouse-type == 'dremio')
+        if: steps.seed-cache.outputs.cache-hit != 'true' && inputs.warehouse-type != 'duckdb' && inputs.warehouse-type != 'trino' && inputs.warehouse-type != 'spark' && inputs.warehouse-type != 'dremio' && (inputs.warehouse-type == 'postgres' || inputs.warehouse-type == 'clickhouse')
         working-directory: ${{ env.E2E_DBT_PROJECT_DIR }}
         run: |
           CACHE_DIR="/tmp/seed-cache-${{ inputs.warehouse-type }}"
@@ -356,12 +356,6 @@ jobs:
                 echo "Waiting for Spark after restart... ($i/60)"; sleep 5
               done
               ;;
-            dremio)
-              for i in $(seq 1 60); do
-                curl -sf http://localhost:9047 > /dev/null && break
-                echo "Waiting for Dremio after restart... ($i/60)"; sleep 5
-              done
-              ;;
             postgres)
               for i in $(seq 1 30); do
                 pg_isready -h localhost -p 5432 > /dev/null 2>&1 && break