Quote string values in arrays? #307

As a followup to larray-project/larray#1165, I wonder whether we should quote them here too.

On one hand it would make those values more obvious in mixed type arrays, which is very important because even though we already have the background color hint, I am pretty sure most of our users wouldn't know what it means. On the other hand, it would look odd when the entire array contain strings, or even an entire column in a "Frame" structure.

FWIW, Pandas does not quote those values:

>>> pd.DataFrame({'num': [1, 2, 3], 's': ['a', 'b', 'c'], 'mixed': [1, "2", "three"]})
   num  s  mixed
0    1  a      1
1    2  b      2
2    3  c  three

I am leaning towards making this the case for object arrays but not string arrays. What I am sure of is that it should be up to the adapter to decide because for some data types (e.g. DirectoryAdapter) it would look REALLY odd/out of place.

For mixed type columns it would clearly be a plus for *frame structures but it might be too expensive to make the difference between object dtype with homogeneous type values vs object dtype with mixed type values. Pandas3 with its "real" string type by default (instead of using object dtype for strings) should help if we want to do it for object but not string columns.