Problem
----

* Given an HTML file with charset "windows-1255". 

* Opening the file from disk detects the encoding correctly.

* Opening a ZIP archive with the same file inside and than opening the
  HTML archive member does not detect the encoding, instead the coding
  system for saving is the default according to M-x
  describe-coding-system.

Attached are two files test.html and test.zip.  Call "emacs -Q test.html
test.zip" and press RET on the archive member to reproduce.