bug#21867: 25.0.50; lossage's log doesn't treat characters read by read-char as separate commands

[Juri Linkov]

>>   C-x C-e                 ;; eval-last-sexp
>>   a                       ;; self-insert-command
>
> This specific output would be wrong, but I guess we could output
> something like:
>
>     C-x C-e                 ;; eval-last-sexp
>     a                       ;; <during eval-last-sexp>

Is it safe to assume that any characters not belonging to
a keysequence bound to a command, are read by the previous
command?  If yes, then I guess one of the patches that Lars
sent (with 'non-command-character') could be adapted
for this output.