Reduce side of bulks when indexing in dev, tests and prod

It seems documents have gotten bigger and are now saturating the
OpenSearch memory, which is rather restricted in our current setup,
leading to failures.

src/main/resources/application.properties

@@ -72,10 +72,10 @@ quarkus.hibernate-search-standalone.elasticsearch.version=opensearch:2.18
 quarkus.elasticsearch.devservices.image-name=opensearch-custom:${maven.version.opensearch}
 quarkus.elasticsearch.devservices.java-opts=${PROD_OPENSEARCH_JAVA_OPTS}
 # Limit parallelism of indexing, because OpenSearch can only handle so many documents in its buffers.
-# This leads to at most 8*20=160 documents being indexed in parallel, which should be plenty
+# This leads to at most 8*12=96 documents being indexed in parallel, which should be plenty
 # given how large our documents can be.
 INDEXING_QUEUE_COUNT=8
-INDEXING_BULK_SIZE=20
+INDEXING_BULK_SIZE=12
 quarkus.hibernate-search-standalone.elasticsearch.indexing.queue-count=${INDEXING_QUEUE_COUNT}
 quarkus.hibernate-search-standalone.elasticsearch.indexing.max-bulk-size=${INDEXING_BULK_SIZE}
 # We need to apply a custom OpenSearch mapping to exclude very large fields from the _source