AWS Rekognition and Google Vision use spaces in multiword tags ("Clock Tower", "Leisure Activities", "black and white"). Magnolia uses underscore ("prairie_chicken"). Let's make Magnolia local image recognition work the same way.

Acceptance criteria:

• In the backend, replace underscore with space
  • when a tag contains multiple words, use the space character as separator

Note: Check if space is allowed in content tags to begin with. Could be a limitation of tagging rather than image recognition.