tabulate: fix categorical and string edge cases

[beingamanforever]

tabulate: categorical and string edge-case fixes

This change improves tabulate to handle categorical and string edge cases correctly, aligning behavior more closely with MATLAB.

Fixed

• Categorical arrays with all values undefined now return zero counts and zero percentages (instead of NaN)
• Categorical arrays with defined but unused categories correctly include zero-count rows
• String arrays with all values missing now return an empty table
• Categorical logic is guarded and only enabled when categorical support is available

Tests

Added focused BISTs covering:

• All-undefined categorical inputs
• Empty categorical arrays with predefined categories
• All-missing string arrays

Tests are automatically skipped when required datatypes are unavailable.

[pr0m1th3as]

See my comment on #351

[beingamanforever]

Updated to keep changes minimal:

• Fixed division-by-zero when all categorical values are undefined (now returns zeros instead of NaNs)
• Removed a duplicated code block in string handling
• No guards added, all existing BISTs preserved, and new tests only cover the fixed edge cases.

Please let me know if you’d prefer these changes split differently or adjusted further.