eloebel/glacier-front-extraction: Release v2.0.0

Major update to the deep learning processing system for automated calving front delineation. The background for this update is the work done in our latest study.

Main changes:

• Training data extended by 423 new glacier fronts, of which 252 are from the Antarctic Peninsula. Please use the updated model-weights.tf.
• Most of the major glaciers (specifically those with frontal widths greater than 5 km) on the Antarctic Peninsula can be called directly from the environmental variability (see new table). This includes the Hektoria-Green-Evans Glacier system, which is processed here with 6 overlapping predictions.
• Recommended use of this software is no longer via containers (Docker, Singularity), but directly via the scripts in the repository (requirements.txt has been updated). Container versions will no longer be updated.