update hightable and implement dataframe v2

[severo]

@platypii do you want to review this one?

I'm not really sure about the dataframe I created for the iceberg files.

Three things to think about:

• how to handle the case where the file contains less than the estimated numRows. In this implementation, I put -1 in the row number (as we only support numbers - it could also be NaN) and undefined in the cells
• I'm not sure how icebergRead works, but I fear it has to fetch from the first row, even if we ask for [1000, 2000]. Is it right? If so, do you have an idea of how we could improve? maybe asking for [0, rowEnd] and cache everything to avoid later fetches (consumes more memory)
• Currently, if we scroll slowly, we send a lot of icebergRead for one or two rows. We could improve by batching by 1000 rows (a bit like VirtualRowGroup in hyperparam-cli)

Re cache: we might want to implement hyparam/hightable#232, and use it here, to simplify the cache management.

[platypii]

Looks good. Honestly I'm not that worried about the iceberg functionality. Eventually I think we will need some support for variable number of rows, but this is good for now 👍