DictEncode a DictEncode

DictEncode signals that a column/array should be dictionary encoded when serialized to the arrow streaming/file format.

The current constructor will happily wrap a DictEncode in another DictEncode.

https://github.com/apache/arrow-julia/blob/v2.2.1/src/arraytypes/dictencoding.jl#L69

Does it make sense to add a no-op constructor `DictEncode(x::DictEncode) = x`?
Right now `Arrow.write` fails to if I run this line twice. I know it's not good practice and it's no big deal, but wanted to bring it up for discussion anyway.

```julia
df.col= Arrow.DictEncode(df.col)
```

Start at src/arraytypes/dictencoding.jl around line 69 and inspect the DictEncode constructor. Reproduce the issue by applying Arrow.DictEncode twice to df.col and then calling Arrow.write; done means repeated wrapping no longer causes Arrow.write to fail.

Written by the indexing model from the issue text.