Compute memory usage of bid pandas dataframes

As discussed in https://github.com/jupyterlab-contrib/jupyterlab-variableInspector/pull/319#pullrequestreview-2279671128

It would be nice to compute the memory_usage of big pandas dataframes. One proposed solution is to do:

```python
sum([
    size_of(dtype) * count
    for dtype, count in x.dtypes.value_counts().items()
]) * len(x)
```