Added delta decoding logic for split index32 and split index 64

[Krmjn09]

The testing if the logic is still pending because I am unable to produce root file for index and splitindex columns

[silverweed]

Thanks! I left a couple of comments.

Also, to produce columns of type Index64/SplitIndex64 you can just create Strings. If the RNTuple is compressed (default), the principal column of a String will be of type SplitIndex64; if you want a non-split column you can create another string field like this:

  auto nonSplitString = std::make_unique<RField<std::string>>("nonSplit");
  nonSplitString->SetColumnRepresentatives({{ENTupleColumnType::kIndex64, ENTupleColumnType::kChar}});
  model->AddField(std::move(nonSplitString));

We don't have a way to automatically create fields with (Split)Index32 types, but you can do the same thing described above and replace kIndex64 with the 32 versions, so you end up with an RNTuple with 4 strings, each using a different column type for its indices.

[Krmjn09]

After testing the code now on strings.root file I can successfully see the values for all 4 types of columns as you can see below

[silverweed]

Very good! Next step will be decoding the zigzag-encoded SplitInt column types.
Also, please open a new PR where you add some testing of this string decoding inside rntuple_test.js (you can add the strings.root RNTuple in the repo for that)