Added summary and total rows to top command.

[data-pup]

This PR makes a few changes to the top sub-command in order to introduce a summary of remaining items when the output is truncated, and a row containing a total of all items. If the retaining size is being used, then these columns will only show "..." in the emitted table.

Other minor changes include changing the name of the number field to max_items, for the sake of consistency with other sub-commands. This has also been changed from an Option<u32> into a plain u32 value. In order to prevent a breaking change from introducing truncation where there previously was none, this defaults to u32::MAX.

I refrained from doing so in this PR, but it seems like a good idea to decide on a consistent default for the maximum number of items to display. diff for example, defaults to 20, while others show 10. Once an --all flag is introduced, we can go about doing this, and update the test cases as well.

Other note: For now, I kept the TableRow struct internal to the text emission function, but this could eventually be refactored out into an entry struct similar to the DiffEntry and MonosEntry structs shortly, assuming the csv and json emission eventually use it as well.

(This relates to #88)

[data-pup]

Rebased this off the new master after merging in #92, due to previous CI errors :)

[data-pup]

CI issue popped up again, taking care of this over in #97 :)

[fitzgen]

Wow! Thanks @data-pup :)

[fitzgen]

I refrained from doing so in this PR, but it seems like a good idea to decide on a consistent default for the maximum number of items to display. diff for example, defaults to 20, while others show 10. Once an --all flag is introduced, we can go about doing this, and update the test cases as well.

+1 -- also don't think we should be afraid of breaking changes for what is 99% a CLI tool and isn't stable yet at that.

[fitzgen]

Other note: For now, I kept the TableRow struct internal to the text emission function, but this could eventually be refactored out into an entry struct similar to the DiffEntry and MonosEntry structs shortly, assuming the csv and json emission eventually use it as well.

Yeah, some of the earlier Emit implementations do a lot more duplicated work in each emit_* method than they should.