is this e.g. due to the verbosity of re-declaring field names every time? or something else? e.g. CSVs handle 2D tables reasonably efficiently, since you just

is this e.g. due to the verbosity of re-declaring field names every time? or something else? e.g. CSVs handle 2D tables reasonably efficiently, since you just specify values: field_a, field_b, field_c 1, 2, 3 3, 4, 5 but I imagine TOML would benefit from something that lets you define nested tables without re-defining the table every time... which probably conflicts with the "obvious" goal. Hmm.