The use case is a Banana dashboard working with a SolrCloud instance/cluster. If we follow default steps we end up using the 'data_driven_schema' in Solr, which makes it easy for it to accept any random data and try index it.
However, the problem is down the line. Banana table widget can't sort on many columns as Solr complains about those fields being multi-valued. In fact, they are not, and unique (checked via admin section), but rather declared multi-valued.
What is the approach to address this? Ideally, without having to specify a complete new schema for a Solr index. Can one have a benefit of flexible fields, but default to non-multi-valued maybe?