Assign to multiple columns in `@transform` (and other applicables)

In DataFrames proper I can do something like

df = DataFrame(point = [(1, 2), (2, 3), (3, 4)])
transform!(df, :point => [:x, :y])
transform!(df, [:x, :y] => ByRow((x, y) -> expensive_computation(x, y)) => [:res1, :res2])

which automatically expands the returned iterable onto two new columns. As far as I can see, this would map to

df = DataFrame(point = [(1, 2), (2, 3), (3, 4)])
@rtransform!(df, [:x, :y]=:point)
@rtransform!(df, [:res1, :res2]=expensive_computation(:x,:y))

in DataFramesMeta, however, this does not seem to be currently possible.
It is possible to fuse these operations with @astable:

df = DataFrame(point = [(1, 2), (2, 3), (3, 4)])
@rtransform!(df, @astable begin
    :x = :point[1]
    :y = :point[2]
    intermediate = expensive_computation(:x, :y)
    :res1 = intermediate[1]
    :res2 = intermediate[2]
end)

but then again, being able to write this as

@rtransform!(df, @astable begin
    :x, :y = :point
    :res1, :res2 = expensive_computation(:x, :y)
end)

Would be much more concise. Is there a reason that this behaviour is not implemented?

Start with the existing @rtransform! and @astable entry points, and compare their behavior with DataFrames' transform! multiple-column assignment shown in the issue. Determine whether the concise @astable assignments can support tuple or iterable expansion to multiple target columns; done means both examples work as described.