Flux: Flattening of stream records with array property in the record type

I am looking to sanitise some data for forecasts made from a point in time, stored in a field as JSON. I have been able to convert this JSON column field into an array type, and have been able to conform to the following signature:

stream[{_time: time, _values: [{_at: time, _value: float}]}]

I intend to "flatten" this stream of arrays into something where I can see forecast convergence, essentially, creating a sort of pivot. But for this, I need to perform some operation to get this signature:

stream[{_time: time, _at: time, _value: float}]

Essentially, for every record in the stream, there will be 0...n records on the output, but they will be within the same time domain before the pivot.

I am looking to use a query/task within Influx to deal with this before taking to an external solution to manage the "flattening" of nested values. The records are also being written by a tool I have limited control over.

Start with the Flux query/task capabilities for iterating an array-valued record and check whether a built-in operation can produce one output record per nested value. Done means transforming stream[{_time: time, _values: [{_at: time, _value: float}]}] into stream[{_time: time, _at: time, _value: float}] while preserving the time fields and supporting zero or more output records.