Always process the freshest record when ordered by key

Hello, I have a use case where I need to process each key sequentially however when processing a record, if new records arrive for that key, I would like to skip over the ones that have already arrives and process the latest one next.

I tried to read and understand the documentation but I believe currently the library does not support this kind of behavior. Am I wrong? If not, is there a not so hacky way to achieve this?

Start by reviewing the documentation and the key-sequential processing and queueing behavior described there. Compare the current behavior with the requested handling of newly arrived records, especially skipping older records for a key. Done means establishing whether the behavior is supported and documenting a supported approach or defining the implementation scope.

Written by the indexing model from the issue text.