Changed deriving of Functor, Foldable, Traversable to fix #7436. Added foldMap to derived Foldable instance.

The derived instances will no longer eta-expand the function. I.e. instead of
    fmap f (Foo a) = Foo (fmap (\x -> f x) a)
we now derive
    fmap f (Foo a) = Foo (fmap f a)

Some superflous lambdas are generated as a result. For example
    data X a = X (a,a)
    fmap f (X x) = (\y -> case y of (a,b) -> (f a, f b)) x
The optimizer should be able to simplify this code, as it is just beta reduction.

The derived Foldable instance now includes foldMap in addition to foldr.

Compiler-generated code can have empty cases

 * Make MatchGroup into a record, and use the record fields

 * Split the type field into two: mg_arg_tys and mg_res_ty
   This makes life much easier for the desugarer when the
   case alterantives are empty

A little bit of this change unavoidably ended up in the preceding
commit about empty case alternatives

The main changes are:
  * Parser accepts empty case alternatives
  * Renamer checks that -XEmptyCase is on in that case
  * (Typechecker is pretty much unchanged.)
  * Desugarer desugars empty case alternatives, esp:
      - Match.matchWrapper and Match.match now accept empty eqns
      - New function matchEmpty deals with the empty case
      - See Note [Empty case alternatives] in Match

This patch contains most of the work, but it's a bit mixed up
with a refactoring of MatchGroup that I did at the same time
(next commit).

See Note [Case elimination: lifted case]:

We used to do case elimination if
        (c) the scrutinee is a variable and 'x' is used strictly
But that changes
    case x of { _ -> error "bad" }
    --> error "bad"
which is very puzzling if 'x' is later bound to (error "good").
Where the order of evaluation is specified (via seq or case)
we should respect it.

c.f. Note [Empty case alternatives] in CoreSyn, which is how
I came across this.

Apparently this fixes the build with dblatex 0.3.4.

It now takes care of adding p to the GhcLibWays, rather than just
complaining that it's missing.

This means that we can use the standard MonadIO class, rather than
needing our own copy.

I was tracking down an error looking like
  Prelude.(!!): index too large
which is very unhelpful.  This patch replaces at least some uses
of (!!) in GHC with getNth, which has a more helpful error
message (with DEBUG anyway)

This bug was making GHC loop when printing external core from test T7239.

This fixes Trac #7541, and is on by default.
Use -fno-warn-duplicate-constraints to switch it off.

This follows the correspondig change in ClsInst

Trac #7536 points out that it's possible for the LHS to *look*
as if it binds variables, but does not acutally do so

         type T a = Int
         type instance F (T a) = a

This patch makes it an error.

See Note [SingI and EvLit] in TcEvidence.

We now have the invariant for a ClsInst that the is_tvs field
is always completely fresh type variables. See
Note [Template tyvars are fresh] in InstEnv.

(Previously we frehened them when extending the instance environment,
but that seems messier because it was an invariant only when the
ClsInst was in an InstEnv.  Moreover, there was an invariant that
thet tyvars of the DFunid in the ClsInst had to match, and I have
removed that invariant altogether; there is no need for it.)

Other changes I made at the same time:

 * Make is_tvs into a *list*, in the right order for the dfun type
   arguments.  This removes the wierd need for the dfun to have the
   same tyvars as the ClsInst template, an invariant I have always
   hated. The cost is that we need to make it a VarSet when matching.
   We could cache an is_tv_set instead.

 * Add a cached is_cls field to the ClsInst, to save fishing
   the Class out of the DFun.  (Renamed is_cls to is_cls_nm.)

 * Make tcSplitDFunTy return the dfun args, not just the *number*
   of dfun args

 * Make InstEnv.instanceHead return just the *head* of the
   instance declaration.  Add instanceSig to return the whole
   thing.

This adds the missing coercions in the constructed evidence for SingI.
Previously we simply passed an integer or a string for the evidence,
which was not quite correct and causes errors when the core lint is
enabled.   This patch corrects this by inserting the necessary
coercions.

This minor refactoring re-attaches Note [Add unfolding for scrutinee].
It had become detached, which led me on a bit of a wild goose
chase.

While I was at it, I made the code work right for the case where
the scrutinee is of form (x |> co); I don't think this is an important
improvement.

I also make simplAlt unconditionally zap occurrence information on
case-alternative binders (see Note [Case alternative occ info]);
it was almost always being zapped and the additional complexity seems
not worth it.

Move the "combine indentical alternatives" transformation *before*
simplifying the alternatives.  For example
     case x of y
        [] -> length y
        (_:_) -> length y }

If we look *post* simplification, since 'y' is used in the
alterantives, the case binders *might* be (see the keep_occ_info test
in Simplify.simplAlt); and hence the combination of the two
alteranatives does not happen.  But if we do it *pre* simplification
there is no problem.

This fixes Trac #7360.