TODO

Implement flagging-app specific data operations such as:
1. Define data manager obj with apply flags behavior. How does that affect dataset and metadata values (actris level, processing_script)?
2. Define metadata answering who did the flagging and quality assurance tests?
Indicate all measurements are assumed to be valid unless checked otherwise (Should this be in the title? this may overlap with the following item.)
When Create Flag is clicked, modify the title to indicate that we are in flagging mode and ROIs can be drawn by dragging.
Update Commit Flag logic: 3. Update recorded flags directory, and add provenance information to each flag (which instrument and channel belongs to).
Record collected flag information initially in a YAML or JSON file. Is this faster than writing directly to the HDF5 file?
Should we actively transfer collected flags by clicking a button? after commit button is pressed, each flag is now stored in an independent json file
Enable some form of chunk storage and visualization from the HDF5 file. Iterate over chunks for faster display versus access time.
1. Do I need to modify DIMA?
2. What is a good chunk size?
3. What Dash component can we use to iterate over the chunks?