[DEV-1122] [feature proposal] Add native support for pandas API

[ObservedObserver]

For now, pygwalker kernel computation uses sql + duckdb for data queries. Another approach might be using the native pandas API for all those computations. Benefits of this implementation include:

• Test and switch to different high-performance dataframe library, like modin, polars.

It would be even better for the community if developers could customize their own query engines.

_DEV-1122