Fix the coverage checker's treatment of existential tyvars
Previously, the pattern-match coverage checker was far too
eager to freshen the names of existentially quantified type
variables, which led to incorrect sets of type constraints that
misled GHC into thinking that certain programs that involve nested
GADT pattern matches were non-exhaustive (when in fact they were).
Now, we generate extra equality constraints in the ConCon case of
the coverage algorithm to ensure that these fresh tyvars align
with existing existential tyvars. See
Note [Coverage checking and existential tyvars] for the full story.
Test Plan: make test TEST="T11984 T14098"
Reviewers: gkaracha, bgamari, simonpj
Reviewed By: simonpj
Subscribers: simonpj, rwbarton, thomie, carter
Differential Revision: https://phabricator.haskell.org/D4434