Chardata moretests

[pp-mo]

Closes #6920
List of issues from there ...

• READ + WRITE
  • here and here** handling of invalid "_Encoding" values
    • DONE IN Chardata first working on feature branch #6975 (already merged) : here and here
  • n-bytes <-> n_chars width translations for all supported encoding types
    • DONE IN Chardata first working on feature branch #6975 (already merged) : here
  • check alternate names for encodings work, e.g. "UTF8" --> "utf-8"
    • unit test in "...unit/fileformats/netcdf/test_bytecoding_datasets.py"
    • integration test in "...integration/netcdf/test_stringdata.py"
    • DONE IN e6bdb53
  • ALL IN 4e956e0 :
    • different encodings on different variables
    • scalar char variable
  • "bad" unicode content can be read OK as character arrays (non-decode loading) : AND written back
    • DONE IN 6c4be38
• READ ONLY
  • non-decoded reading (using DECODE_TO_STRINGS_ON_READ)
    • ( should produce cubes/coords with a "string dimension", and "S1" type character arrays as content )
    • DONE IN dd5330f
  • ALL IN 2c276a8 :
    • read unicode (utf-8) data with no _Encoding should work OK (-> utf-8 default applied)
    • errors when unicode data read with WRONG encoding
  • check we can still handle variable-length strings : cf Handle NetCDF variable length strings (and other VLen types) #6340
    • ALREADY COVERED by these tests
• WRITE ONLY
  • ALL IN 2c276a8 :
  • error when unicode data + no _Encoding
  • error/warning when encoded strings --> over-length (from n_chars -> n_bytes=dim translation)
  • error when cube/coord data has type "Sxx" other than "S1"

Plus:

• remove "PERSIST_TESTFILES" and other debug in test_stringdata / tests_bytecoding_datasets
  • DONE IN 5d5bedc

Change to pre-commit config merged from main
(into feature-branch)

[pp-mo]

OK I'm finally happy with this!

In my mind, this confirms that the new chardata support facility is "safe to merge"™️
- because above all, we do want to get rid of a feature branch ASAP

Some issues remains in the original plan though.
For reference :

• Check coverage of chardata handling code + purge unreachable #6921
• Review all legacy issues which should be fixed by new chardata support #6922
• Resolve remaining choices/questions for chardata handling #6923
• Complete documentation of chardata handling #6924

[trexfeathers]

• Read through the total changes
• Has this delivered on each checklist item?
• Has this removed all TODO and/or commented code?

[pp-mo]

Weird problem here where GitHub isn't picking up a commit added to the fork branch.
Close+reopen to see if it fixes..

[pp-mo]

OK I think I covered your comments now?

[trexfeathers]

Thanks @pp-mo 👍