Update Row repr to be consistent with Table repr - #4066

To me it's not terribly natural to represent a single row as though it were a single-row table with all the column headings, etc. But I could see the argument for it.

@embray - I find the current repr (and str) completely useless:

>>> t = simple_table(cols=20)
>>> t[1]
<Row 1 of table
 values=(2, 2.0, 'd', 5, 5.0, 'g', 8, 8.0, 'j', 11, 11.0, 'm', 14, 14.0, 'p', 
17, 17.0, 's', 20, 20.0)
 dtype=[('a', '<i4'), ('b', '<f4'), ('c', 'S1'), ('d', '<i4'), ('e', '<f4'), ('f', 'S1'), 
('g', '<i4'), ('h', '<f4'), ('i', 'S1'), ('j', '<i4'), ('k', '<f4'), ('l', 'S1'), ('m', '<i4'), 
('n', '<f4'), ('o', 'S1'), ('p', '<i4'), ('q', '<f4'), ('r', 'S1'), ('s', '<i4'), ('t', '<f4')]>

From a coding (and thinking) perspective the new look was the simplest and fastest, but if you have any other ideas I'm happy to listen.

I like the fact that you get units now.

My personal preference is still for something more tuple-like, with units inline with the numeric values. But I'll give you that--this is definitely better than the existing situation 👍

That's interesting and definitely makes me think about it. I still worry that once you get beyond a handful of columns that it gets difficult to represent the name, type, value, and unit in a way that lets the user quickly associate all those for a single column.

Yeah, you're probably right. I don't have any better ideas.