Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Austin Seipp <austin@well-typed.com>

(We might want to revisit this.)

More often than not the output of the performance tests is in the way,
rather than helping. This allows the use of `make SKIP_PERF_TESTS=YES`
to skip these tests. Fixes #8413

Splices are now run in the renamer instead of the type checker, so they
only get run once.

Because splices are now run in the renamer, we do not get the same error context
as we would when running in the type checker. In most cases we get less context,
and in some cases I have added additional context. Error messages should at
least tell the user that an error occurred in a splice; dropping context beyond
that point is not judged a great loss.

Note that we may now report only one error when multiple errors were reported
before because splices are now run in the renamer.

From the new Template Haskell proposal at
http://hackage.haskell.org/trac/ghc/blog/Template%20Haskell%20Proposal

  A declaration group is the chunk of declarations created by a top-level
  declaration splice, plus those following it, down to but not including the
  next top-level declaration splice. Then the type environment seen by reify
  includes all the declaration up to the end of the immediately preceding
  declaration block, but no more.

This change adds '$(return [])' where necessary to allow following declarations
to see (module-local) top-level definitions.

GHC.TypeLits does not export `Sing` anymore, so we just define it
locally.

There is still one more test that needs fixing:

   indexed-types/should_fail  T7786 [stderr mismatch] (normal)

I need to understand what is going on there, as it appears to be
using the `Sing` constructors a bunch.

As a result of fixing Trac #8384

Authored-by: Carter Tazio Schonwald <carter.schonwald@gmail.com>
Signed-off-by: Austin Seipp <austin@well-typed.com>

Issue #8393

Signed-off-by: Austin Seipp <austin@well-typed.com>

Issue #8393

Signed-off-by: Austin Seipp <austin@well-typed.com>

Issue #8393

Signed-off-by: Austin Seipp <austin@well-typed.com>

following
  a) suppressing kind foralls and arguments
  b) better fundep error messages