Index.slice_locs Missing Argument Only Raises with NA Data

[WillAyd]

xref #20915 perhaps another issue here - I can't tell if this is by design or not:

>>> pd.__version__
'0.23.0.dev0+824.gc4da79b5b.dirty'

>>> index = pd.Index([1., 2.])
>>> index.slice_locs(1.5)
(1, 2)

>>> index = pd.Index([1., 2., np.nan])
>>> index.slice_locs(1.5)
pandas._libs.hashtable.Float64HashTable.get_item
    raise KeyError(val)
KeyError: 1.5

[phofl]

cc @WillAyd The raise in the second seems to be ok, but missleading error message I think? slice_locs only works when Index is monotonic, not the case with nan.

But I could not find any information if this should raise in general when Key is not present in Index?