Flood detection with S1-GRD and S2-L2A - #78

Summary

Contributes to #13

• Adds an openEO conversion of Visualizing Floods by combining optical and radar satellite imagery and classifying stale water bodies as flood. Flood is rendered in blue, overlaid on grayscale S2 scene.
• This is a single-date flood visualization. The presence of flood is clarified with the scenes in pre- and post-flood timings.
• The exported UDP runs successfully in openEO browser, connected to CDSE. Parameters for spatial and temporal extent, as well as for water/flood threshold, and RGB brightness are configurable.
• Flood detection UDPs pass APEx QA tests. Added github workflow for test transparency.
• The process graph has been verified against a flooding area in the UK before, during, and after flood

Test

• Confirm if the notebook runs
• Check for other locations where you know big flooding has recently happened (not a must)
• See if github action on QA tests for flood_detection are passed (other failing tests unrelated to flood detection can be ignored)

Issues

• .github/workflows/run_apex_qa_tests.yml requires access tokens to clone developmentseed/apex_algorithm where the QA tests live. Still based on my personal token, will move to a more sustainable key use later.
• The github workflow will test all existing algorithms in this repository to ensure changes in one algorithm doesn't lead to changes in others. Expect failure as some algorithms may not be aligned to the format approved by APEx