[m-rev.] for review: do occurs checks on the parse tree

[Zoltan Somogyi]

For review by anyone. I am particularly looking for feedback on the
issue of warnings for ok code, such as worst_purity(P1, P2) \= P2.
Can people think of a better way to handle them? All the ways I can
think of to avoid warning about such cases are far too expensive
in compiler execution time compared to the rarity of the misleading
warnings they would avoid.

This addresses Github issue 77.

Zoltan.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Log.occurs
Type: application/octet-stream
Size: 1656 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20191001/ebba1aa6/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: DIFF.occurs
Type: application/octet-stream
Size: 27332 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20191001/ebba1aa6/attachment-0003.obj>