[m-rev.] for review: improve actual/expected types in diagnostics

[Zoltan Somogyi]

For review by anyone. I would also appreciate any real
code examples of similar one actual vs two+ expected
mismatches where at least one of the expectation sources
is not a predicate, though I will use fake ones if I have to.

Zoltan.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Log.ae
Type: application/octet-stream
Size: 3287 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20221016/5555fe38/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: DIFF.ae
Type: application/octet-stream
Size: 78011 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20221016/5555fe38/attachment-0003.obj>