[BUG] Ambiguous indexing+ due to duplicate categorical axis entries

[alexander-held]

Describe the bug

It is possible to define a categorical axis with repeated entries in boost_histogram:

import boost_histogram as bh

h = bh.Histogram(bh.axis.Regular(1, 0, 1), bh.axis.StrCategory(["a", "a"]))
h.fill(0.5, "a")

print(f" 0 : {h[:, 0].values()}")  #  0 : [1.]
print(f" 1 : {h[:, 1].values()}")  #  1 : [0.]

When accessing this with indexing+ via "a" in hist, the first match is returned (at least in this example):

import hist

h = hist.Hist.new.Regular(1, 0, 1).StrCat(["a", "a"], name="a").Double()
h.fill(0.5, a="a")

print(f"\"a\": {h[:, "a"].values()}")  # "a": [1.]
print(f" 0 : {h[:, 0].values()}")  #  0 : [1.]
print(f" 1 : {h[:, 1].values()}")  #  1 : [0.]

I do not know if there are use cases where duplicate categories would be needed. I ran into this during debugging. Otherwise I wonder if this should be explicitly disallowed at boost_histogram or hist level to avoid downstream ambiguity for uhi, or if some kind of warning should be raised here.

Steps to reproduce

see above