uncommitted history has unexpected implications for bundle naming

[kyocum]

The uncommitted history currently maintains a maximum of N uncommitted bundles with the same human name. This can have the unintended consequence of deleting bundles needed by downstream tasks if users do not assign unique human names to the bundles. Note that, if the user does not set the human name in pipe_requires, then this shouldn't be an issue. But what we need to determine is a.) whether an uncommitted history is important to keep as a feature and b.) if so, is there a more intuitive mode of operation.

[kyocum]

Uncommitted history has been retired as a feature.