Content is an analyzed text field, but a single-term KQL query on it was
compiled to a term query, which matches the raw value against the tokenized
content and never hits. "Content:https://opencloud.eu/" (tokenized to
[https, opencloud.eu, ...]) returned nothing; "Content:alan@" likewise.
Route full-text fields to match_phrase so the query is analyzed the same way
as the indexed content. Fixes the OpenSearch content-search acceptance
scenarios (contentSearch.feature search by different content types).
Dominik Schmidt
054337f4fd