Provide a fallback tree watering feature for those who don't have Semantics API yet

[MrCsabaToth]

Since the Semantics API is not publicly available yet, we could provide a fallback tree watering possibly based on Google Vision API annotations:

• https://cloud.google.com/vision/docs/object-localizer
• https://cloud.google.com/vision/docs/reference/rest/v1p2beta1/images/annotate#block