Align MicroSeries.nullify_weights to the Series index

[vahid-ahmadi]

Closes #301

Problem

nullify_weights built its all-ones weight Series without an index, so it carried a default RangeIndex. Weighted operations are label-aligned (self.multiply(self.weights) in sum()/weight()), so on any non-default index the multiply produced all-NaN and every aggregation collapsed to zero:

s = MicroSeries([1, 2, 3], index=[10, 11, 12], weights=[1, 2, 3])
s.nullify_weights()
s.sum()      # -> 0.0, expected 6.0

With a default index it happens to work, which is why nothing caught it.

This is the same bug class as #283, which added index=self.index to set_weights (with a comment explaining the label-alignment trap) but left nullify_weights on the old pattern.

Change

Pass index=self.index, plus a comment matching the one in set_weights, and a regression test covering sum, mean, and the resulting weight index.

Verification

• New test fails before the change (0.0 == 6 assertion), passes after
• make test — 63 passed on pandas 3.0.5 and pandas 2.3.3
• make lint — passes

Note: the lint job will be red until #307 merges (unrelated ruff 0.16 issue on main).