How to approach add_nested with mismatched index

[dougbrn]

add_nested currently works as a left join from the perspective of the base layer, as shown below:

# unmatched nested objects are removed
>>> base = NestedFrame({"a":[1,2,3]}, index=[0,1,2])
>>> nested = NestedFrame({"b":[1,2,3,4]}, index=[0,1,2,3])

>>> base.add_nested(nested, "nested")
	a	nested
0	1	b 0 1
1	2	b 0 2
2	3	b 0 3

# unmatched base rows are kept
>>> base = NestedFrame({"a":[1,2,3]}, index=[0,1,2])
>>> nested = NestedFrame({"b":[1,2,3,4]}, index=[0,1,3,3]) # no match for index=2

>>> base.add_nested(nested, "nested")
	a	nested
0	1	b 0 1
1	2	b 0 2
2	3	None

Is this the best default behavior? Should we add various join options to the output of this operation? For time-domain purposes, I can see value in keeping nested lightcurves even when no object information is present

[dougbrn]

This has been implemented on the nested-dask side as a choice that defaults to an outer join, we should follow suit here:

[hombit]

Should be fixed when #115 and #116 are fixed