I wavered on proposing this for a few days, and am leaning towards this idea.
It would be useful for those who are dealing with expensive value generators to query a region and see if an active dogpile lock exists for a given key. (which would then need to be implemented on the backends as well, returning a boolean or perhaps the seconds remaining)
the rationale is that expensive/long-running generators still result in a pileup of processes that are just waiting for a bottleneck to passes. in some known instances, you may want to handle this.
There are other ways around this, I'm using a few myself. This just seems like it would be useful.