This gives csv.py access to a hidden feature in our table renderer used
by some of the other scripts: fields that affect by-field grouping, but
aren't actually printed.
For example, this prevents summing same named functions in different
files, but only shows the function name in the table render:
$ ./scripts/csv.py lfs.code.csv -bfile -bfunction -lfunction
function size
lfs_alloc 398
lfs_alloc_discard 31
lfs_alloc_findfree 77
...
This is especially useful when enumerating results. For example, this
prevents any summing without extra table noise:
$ ./scripts/csv.py lfs.code.csv -i -bfunction -fsize -lfunction
function size
lfs_alloc 398
lfs_alloc_discard 31
lfs_alloc_findfree 77
...
I also tweaked -b/--by field defaults a bit to account to
enumerate/label fields a bit better.
b2768becaa