feat: add TLeafC - string - writing support

[ioanaif]

No description provided.

[jpivarski]

                                          4 bytes          4 bytes
┌─────────┬────────────────────────────────┬───┬────────────┬───┐
│ TKey    │ content                        │ X │ offsets    │ x │
└─────────┴────────────────────────────────┴───┴────────────┴───┘
          │←        fLast - fKeylen       →│                    │
          │                                                     │
          │←                       fObjlen                     →│

These offsets are measured in bytes, starting from the beginning of the TKey.

They do not include the last, so we have to get the last offset from fLast (in the TKey header).

As for what the two "4 bytes" X values are, see the jagged array writer.

Example to follow:

>>> import uproot, skhep_testdata
>>> uproot.open(skhep_testdata.data_path("uproot-simple.root"))["tree"]["three"].array()
<Array ['uno', 'dos', 'tres', 'quatro'] type='4 * string'>

[jpivarski]

This just needs one more test to pass. See below!

[jpivarski]

:)

[jpivarski]

Fantastic! It's great to see this working!

I added some tests of extreme situations, to make sure that fLen is increasing as it should be before and after write_anew, that length-zero string arrays can be written, and that ROOT is still okay with all of these cases. (I had to guess that fLen should be 0 when there are no strings. I could be wrong and maybe it ought to be 1. ROOT didn't complain, though.)

If you have any counter-edits, go ahead and apply them, otherwise, feel free to squash-and-merge.