Support duplicated columns (maybe?)

We currently don't support duplicated columns, introducing support is not completely trivial since we would have to handle them properly in column projections. One example to explain the difficulties:

```
pdf = pd.DataFrame([[1, 2], [3, 4]], columns=list("aa"))
print(pdf["a"])

print(pdf[["a", "a"]])
```

```
0 1 2
1 3 4
a a a a
0 1 2 1 2
1 3 4 3 4
```

Selecting duplicated columns once returns all of them, but having them n times in the indexer duplicates the columns also n times, so we would have to be very careful to keep the count consistent. Other things also depend on set comparisons, which makes this even harder.

I don#t think that this is a very important feature for now, so I don't intend to prioritise this anytime soon

Start by reproducing the duplicated-column behavior in the pandas example from the issue, including both single-name and repeated-name selection. Trace how column projections and set comparisons handle the duplicated labels, then define tests that verify column counts remain consistent for each selection form. Done means duplicated columns are supported without changing the demonstrated selection semantics.

Written by the indexing model from the issue text.