Set Tree Operation

[BenGalewsky]

Problem

There is no way to specify which tree you want an uproot transformer to operate on

Approach

The tree name needs to be appended to the first Call in the AST. It has to be appended along with a filename (which doesn't seem to be used anywhere).

The set_tree operation needs to clone the AST so you can do something like:

q = ds.Select(lambda e: {'el_pt': e['el_pt']})\
    .set_result_format(ResultFormat.parquet)

nominal = q.set_tree("nominal")
mmt_2016 = q.set_tree("mmt_2016")

print(nominal.as_files())
print(mmt_201.as_files())

Cloning the AST was a bit tricky since there is a reference to the FuncADLDataset in a private property, _eds_object.
I wrote a custom __deepcopy__ method to carefully avoid cloning the open file pointer to the cache DB.

I don't think anyone uses _eds_object and we should consider dropping it from the func_adl library

[gordonwatts]

I have no idea what the _eds_object is - it might have been used in func_adl_servicex, but that won't matter any longer. So yeah - this is for some technical debt clean up!

[gordonwatts]

Are there no tests? :-)

[gordonwatts]

Looks cool - I guess I'd expect at least one test - these deep copy things are tricky and if we add something to a event data set, this could cause something funny happening downstream.