[m-rev.] for review: decide each type's representation just once

[Zoltan Somogyi]

Peter, can you please review this, preferably sometime soon?

The parts of the diff concerned with automatically comparing
the output of the new algorithm with the output of the old
are not intended to be committed. I included them both to show
how I did that comparison, and to keep the functionality in case
any changes in response to review comments require
retesting.

Zoltan.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Log.repn
Type: application/octet-stream
Size: 1860 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20180215/daf13699/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: DIFF.repn
Type: application/octet-stream
Size: 92099 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20180215/daf13699/attachment-0003.obj>