[Groff] New request `trin' and asciify

[Werner LEMBERG]

As already mentioned on this list, I have removed almost all `charXXX'
entities from the font definition files (and I will continue).

Right now I discovered another dependency between input and output
encoding.  For example, in latin1.tmac you can find (after macro
expansion)

  .tr \[char229]\[oa]

which maps input code 0xE5 to `a with ring above', disabling the
asciify mechanism.  With the charXXX entries in the font files, this
translation wasn't necessary, and .asciify could map char229 back to
`a with ring above'.

I've added some lines of code to fix this.  `trin' is the same as `tr'
except that .asciify will use the original input code.



    Werner