This report is a followup to bug#35721[1] focussing only on the
alternating composition of Arabic characters when editing seemingly
unrelated parts of the buffer.

[1]: https://debbugs.gnu.org/35721

Observe:

0. emacs -Q
1. C-u C-\ arabic RET
2. a ; C-a C-u C-x =

--8<---------------cut here---------------start------------->8---
             position: 146 of 147 (99%), column: 0
            character: ش‎ (displayed as ش‎) (codepoint 1588, #o3064, #x634)
              charset: unicode (Unicode (ISO10646))
code point in charset: 0x0634
               script: arabic
               syntax: w 	which means: word
             category: .:Base, R:Right-to-left (strong), b:Arabic
             to input: type "a" with arabic input method
          buffer code: #xD8 #xB4
            file code: #xD8 #xB4 (encoded by coding system utf-8-unix)
              display: composed to form "ش" (see below)

Composed using this font:
  xft:-PfEd-DejaVu Sans Mono-normal-normal-normal-*-15-*-*-*-m-0-iso10646-1
by these glyphs:
  [0 0 0 3186 9 -1 9 9 1 nil]

Character code properties: customize what to show
  name: ARABIC LETTER SHEEN
  general-category: Lo (Letter, Other)
  decomposition: (1588) ('ش')

There are text properties here:
  fontified            nil
--8<---------------cut here---------------end--------------->8---

3. C-e RET

The sheen is correctly shaped in its initial form: