Add hasUpperBound and hasLowerBound

This is useful for performing PVP bounds checking, especially for
tooling that wants to automate such dependency bounds. A demonstration
of this being used in practice is the --pvp-bounds flag for stack:

https://github.com/commercialhaskell/stack/commit/c774f9905b3b112f58fc3789dfb9c26389d33b24

Lower memory usage during index file parsing #2396

The parsePackageIndex wasn't streaming because resulting list was
hidden in two constructors: Either and tuple. By removing that error
semantics changed in updatePackageIndexCacheFile.

The updatePackageIndexCacheFile problem was sequence call that evaluated
whole list made by parsePackageIndex. By using lazySequence any error
during index file parsing will be raised during writing to cache file.

Also removed reacPackageIndexFile, as it wasn't used anywhere as far as
I can tell. If this is a mistake I'll include it again.

Revert "Exit with an error if $CABAL_SANDBOX_CONFIG does not point to extant file"

Call addInfoForKnownRepos in initialSavedConfig

We called addInfoForKnownRepos when _parsing_ a file, but not when creating the
initial file. This meant that when we created an initial file, we would end up
with a different internal configuration (which would not have
addInfoForKnownRepos applied) than if we wrote out and then read back that same
initial file (which would).

This reverts commit 3488f593.

Bump the upper bound on 'process'.

Extend other-extensions documentation

Change "XXX" markers to "TODO"

Don't print ExitCode in topHandlerWith

Closes #2807.

Fixes #2805.

Add Bash script for starting development

use a liberal import of Control.Exception to grab the new ErrorCall pattern in base

Use the sandbox config for "cabal get"

If I try to use a local cabal-install repo cache by adding a
"remote-repo-cache: <dir>" setting to the local cabal.config, this is
respected by "cabal update" but then ignored by "cabal get", which seems
wrong.  This respects the local cabal.config in "cabal get" too.  Maybe
all commands should do this?

Remove maybeDecompress_plain property

Fix #2794: Generate ascii data for plain maybeDecompress property

[ci skip]

Resolve #2772: generate database flags (may omit global db)

Warn when constraints or preferences refer to unknown packages (fixes #2719)

Don't strip libraries in build directory

`cabal install` and `cabal configure` warn when any constraint or preference
names a package that is not in the source package index or installed package
index.

Two changes:

  * strip static libraries (.a files) at install/copy time, instead of
    at build time. Note that Cabal strips shared libraries (.so files)
    and executables at install/copy time also.

    This is needed for GHC, because it uses Cabal (via ghc-cabal copy)
    to install the boot libraries, and relies on Cabal to take of
    library stripping. Currently .so files indeed get properly stripped,
    but .a files do not.

    (Stripping static libraries at build time was introduced in
    60409cb9, with the explanation:
    "Call 'stripLib' from createLibArchive so that it's done only
    once.")

    This bug (if you want to call it that) partially explains the
    difference in size of a GHC installation before and after manual
    stripping, reported in
    https://github.com/haskell/cabal/issues/1622#issuecomment-62076817
    (The other parts are that the GHC build system currently doesn't
    strip executables on installation:
    https://ghc.haskell.org/trac/ghc/ticket/9087, and neither are the
    RTS library stripped, since they don't go through Cabal).

  * never strip the copy of a library in the build directory. Only
    strip the copy that is installed. Note that Cabal never strips
    executables in the build directory either.

    This might speed up compilation of a package under development,
    since stripping won't be performed for every 'cabal build'. It is
    also consistent with the GNU coding standards for 'install-strip':

      "install-strip should not strip the executables in the build
      directory which are being copied for installation. It should only
      strip the copies that are installed."

    Source:
    http://www.gnu.org/prep/standards/html_node/Standard-Targets.html

When running the `strip` command (without the -D option) on an installed
.a file, `strip` will write the current time to all file headers in the
.a file. Update the DeterministicAr test to allow this.

The DeterministicAr test is supposed to test for regressions in
relinking avoidance. It is sufficient to do so in the build directory
only.

Remove dead code