Variables View: For data frame columns, consider showing str representation rather than individual values

[jjallaire]

Consider the bulldozer blue book data frame (details on downloading and reading it are here: #966)

The environment pane currently shows up to 100 individual values when you expand this column:

However, there is a print method for the column that IMO provides a much more useful summary (R data frames also support this behavior):

I think that in the Environment pane we should have the notion of data types that terminate in a textual summary (rather than just recursing into long displays of individual scalars)

[jjallaire]

Here's a related example for a variable that is an array of > 400,000 booleans. We show the first 100 values (all True):

Whereas the describe() method shows a more useful summary:

I am hopeful that numpy, pandas, torch, etc. all have ways of inspecting data at a higher level that we can leverage in the environment pane (note that b/c these are all drill downs they don't need to be computed eagerly).

[jjallaire]

Another example is our treatment of Matrices. Here is the Environment pane view (nested lists):

Whereas here is the print() view:

[wesm]

Aside: now that we have the data explorer code to power summary statistics, fulfilling summary requests for the variables pane should be reasonable straightforward, once we have settled on what the new UI treatment is going to look like.