Exchange Section - Categories for processes

[m-mohr]

This issue is capturing some ideas for exchanging processes and categorizing them. This is not necessarily meant for the Hub itself, but for the place where we implement a process exchange.

Processes should be analyzed on submission and submitters should be made aware of the categories and implications during submission.

1. Bound to a back-end
   The process contains back-end specific details, e.g. collection id, band names, file paths, batch job result IDs, save_result options etc. Therefore it usually also contains a load_collection/result and a save_result. These processes can only be run at a specific back-end and thus the submitter must also provide the back-end URL (and API version?) so that people can actually reproduce it.
2. Parametrized data-cube operations
   The process contains parameters for back-end specific details (see above) and has at least one input parameter that accepts a data cube. It should not contain load_* and save_* processes. Some other processes may also be excluded such as the new preprocessing steps as these are often also very back-end specific.
3. Parametrized 'callbacks'
   The process contains parameters for back-end specific details (see above) and has NO parameters that accepts a data cube or any process working on data cubes, so it (usually) works on arrays or scalars.

For 2 and 3 all processes should have well-defined process metadata, so at least id, parameters, return value, summary and description. 2 and 3 can be promoted across back-ends, but 1 should not be as visible as these are more like examples and thus should only be listed at a back-end specific page.