I decided to do the needed rearrangement of frames in this issue, leaving revision of widgets other than the sample to #24776.

PR3960 initially puts the following in the sample box.  I believe it should cover most Python users.

<ASCII/Latin1>
AaBbCcDdEeFfGgHhIiJj
1234567890#:+=(){}[]
¡¢£¥§©«®¶½ÀÁÂÃÄÅÇÐØß

<IPA,Greek,Cyrillic>
ɐɕɘɞɟɤɫɮɰɷɻʁʃʆʎʞʢʫʭʯ
ΑαΒβΓγΔδΕεΖζΗηΘθΙιΚκ
БбДдЖжПпФфЧчЪъЭэѠѤѬӜ

<Hebrew, Arabic>
אבגדהוזחטיךכלםמןנסעף
ابجدهوزحطي٠١٢٣٤٥٦٧٨٩

<Devanagari>
०१२३४५६७८९अआइईउऊएऐओऔ
कगङचजञतदनपबमयरलवशसह॥

<East Asian>
〇一二三四五六七八九
汉字漢字人木火土金水
ᅡᅦᅩᆨᆫ가결걵곴극
あいうえおかさたなま

The new help message explains

Font sample: This shows what a selection of BMP unicode characters look
like for the current font selection.  If a font face does not define a
character, Tk attempts to find another font that does.  Substitute
glyphs will not necessarily have the same size as the font selected.
Hebrew and Arabic letters should display right to left, starting with
alef, \u05d0 and \u0627.  Arabic numerals display left to right.  The
East Asian samples are Chinese digits followed by Chinese Hanzi, Korean
Hangul, and Japanese Hiragana.

Except for Chinese, I intend for the samples to be meaningless subsequences of the respective character set.  I will try to get comments from others for the Indian, Korean, and Japanese samples.

The Chinese reads 'hanzi hanzi person wood fire earth metal water' with 'hanzi' repeated in simplified and traditional character variations.  Louie, could this be controversial in any way?