feat: use new dask methods

[agoose77]

Leverages dask-contrib/dask-awkward#409

[lgray]

@agoose77 this is in the right direction but there's one big problem in that now iterated __getattr__ seems to not be supported:

repro:

from coffea.nanoevents import NanoEventsFactory
events = NanoEventsFactory.from_root(
    "tests/samples/nano_dy.root:Events",
).events()

events.Jet.matched_muons.matched_jet

now fails with:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: 'function' object has no attribute 'matched_jet'

[lgray]

@agoose77

Still running into some regressions:

pytest tests/test_nanoevents_pfnano.py tests/test_nanoevents_physlite.py

results in a couple failed tests with:

TypeError: meta must be an instance of an Awkward Array, not <class 'dask_awkward.lib.core.Array'>.

and

NotImplementedError: 

Which were not encountered before!

[lgray]

@agoose77 ok this last edit gets rid of all the regressions on my side.

[agoose77]

OK, let's look to merging this in dask-awkward, then :)

[lgray]

@agoose77 is there any way we can make it so I don't have to repeat functions like metadata that access awkward-array information but don't need dask? That seems rather pesky.

[agoose77]

@lgray sure! Let me think about it.

[agoose77]

@lgray try this (with new dask-awkward in my branch)

[lgray]

all green in local tests.

[lgray]

Once all that gets merged I'll finally switch nanoevents to use transient attrs and can get rid of that nasty copyreg I use right now. Woo.