.metric_table() and .nearest() don't work in certain cases for delayed nanoevents when return_combinations and return_metric are set to True respectively

[ikrommyd]

Describe the bug
This issue is just for tracking since @lgray said it's almost certainly the fact that "the behavior catching for dask awkward only assumes one output."
The methods .metric_table() and .nearest() don't work in delayed nanoevents when return_combinations and return_metric are set to True respectively and fail with TypeError: meta must be an instance of an Awkward Array, not <class 'tuple'>.

To Reproduce

from coffea.nanoevents import NanoEventsFactory

eagerevents = NanoEventsFactory.from_root(
    {
        "https://github.com/CoffeaTeam/coffea/raw/master/tests/samples/nano_dy.root": "Events"
    },
    delayed=False,
).events()

daskevents = NanoEventsFactory.from_root(
    {
        "https://github.com/CoffeaTeam/coffea/raw/master/tests/samples/nano_dy.root": "Events"
    },
    delayed=True,
).events()

print(eagerevents.Electron.metric_table(eagerevents.TrigObj, return_combinations=True))
print(eagerevents.Electron.nearest(eagerevents.TrigObj, return_metric=True))

print(daskevents.Electron.metric_table(daskevents.TrigObj, return_combinations=True))
print(daskevents.Electron.nearest(daskevents.TrigObj, return_metric=True))

The first two print statements work as expected.
The other two fail with TypeError: meta must be an instance of an Awkward Array, not <class 'tuple'>.

Expected behavior
They should all work.

[ikrommyd]

@lgray please dm me on slack when you get the chance like you said in the awkward-dask meeting

[ikrommyd]

This is partially fixed by #925 but only works with dask_awkward column optimization off. Waiting for upstream dask_awkward fixes to completely close this one.

[lgray]

Fixed by dask-contrib/dask-awkward#430