First off, they mention that when the initial design decision was made, a similar operation cost ~$160, which is tenable for an operation that only happens once in a while. This is in fact a case of them getting bitten by the pricing structure changing after a reasonable design decision (at the time) was implemented.
Secondly, they mention that this is part of a larger issue: "In our most common case we might have to add and delete a couple items to the list property every once in a while. That would still cost us well over $1,000 each time. Most of the reasons for this type of data in our product is to compensate for the fact that there isn't full text search yet. I know they are beta testing full text, but I'm still worried that that also might be too expensive per write."
This is a real problem that GAE needs to solve.
Finally, their problem doesn't seem to be that they need arbitrary precision, its that they seem to need fast location centric queries of a large database.
Geoboxes allow you to solve this problem correctly (and quickly), returning the results in the database that are closest to you. Matching on a geohash can end up serving the incorrect data unless you resort to hacks involving a number of queries.