Fix assertion in prim_tty

The reader updates the Unicode state right before checking the input's encoding, but the assertion that encoding errors are only possible when using utf8 was done against the original state.

lib/kernel/src/prim_tty.erl

@@ -544,7 +544,7 @@ reader_loop(TTY, Parent, SignalRef, ReaderRef, FromEnc, Acc) ->
                             {error, B, Error} ->
                                 %% We should only be able to get incorrect encoded data when
                                 %% using utf8
-                                FromEnc = utf8,
+                                UpdatedFromEnc = utf8,
                                 Parent ! {self(), set_unicode_state, false},
                                 receive
                                     {set_unicode_state, false} ->