[m-rev.] for review: generate mmakefile fragments using ONLY the new approach

[Zoltan Somogyi]

The diff itself is boring and probably does not need review,
but the idea of deleting the old code handgenerating mmakefiles does.

The diff changing over to the new system of creating mmakefile fragments
by creating them as data structures and then writing them out was committed
about three weeks ago. I have been using an installed compiler with the
attached diff in it for about one week, without any problems. Did anyone
else have any mmakefile-related problems in the last three weeks, or does
anyone else have any other objections to committing this diff?

Zoltan.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Log.mm2
Type: application/octet-stream
Size: 416 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20170718/541e3bf4/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: DIFF.mm2
Type: application/octet-stream
Size: 120501 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20170718/541e3bf4/attachment-0003.obj>