ci: Speed up build by using pre-built image from previous stage for `…

…make dev`-like check (#11163)

See https://openfoodfacts.slack.com/archives/C02LDQDDD/p1734814872519649
for discussion.

Pro: Speeds up the `make dev` check by a lot, and it use less disk
space, reducing the risk of the build agent running out of space.
Con: While similar, it doesn't run `make build` as part of the `make
dev_no_build` check, making it not 100% identical to the local `make
dev`

.github/workflows/pull_request.yml

@@ -223,7 +223,7 @@ jobs:
         echo "export USER_GID=$(id -g)" >> .envrc
     - name: Test make dev
       run: |
-        make DOCKER_LOCAL_DATA="$(pwd)" SKIP_SAMPLE_IMAGES=1 dev
+        make DOCKER_LOCAL_DATA="$(pwd)" SKIP_SAMPLE_IMAGES=1 dev_no_build
         make status
     - name: Test all is running
       run: make livecheck || ( tail -n 300 logs/apache2/*error*log; docker compose logs; false )

Makefile

@@ -109,6 +109,13 @@ dev: hello build init_backend _up import_sample_data create_mongodb_indexes refr
 	@echo "🥫 You should be able to access your local install of Open Food Facts at http://world.openfoodfacts.localhost/"
 	@echo "🥫 You have around 100 test products. Please run 'make import_prod_data' if you want a full production dump (~2M products)."
 
+#-------#
+# CI    #
+#-------#
+dev_no_build: hello init_backend _up import_sample_data create_mongodb_indexes refresh_product_tags
+	@echo "🥫 You should be able to access your local install of Open Food Facts at http://world.openfoodfacts.localhost/"
+	@echo "🥫 You have around 100 test products. Please run 'make import_prod_data' if you want a full production dump (~2M products)."
+
 edit_etc_hosts:
 	@grep -qxF -- "${HOSTS}" /etc/hosts || echo "${HOSTS}" >> /etc/hosts