Investigate integration into ghci

[judah]

Haskeline is fully featured enough to make it worth trying to replace editline in ghci.

I expect this change to make ghci somewhat simpler; for example:

• ghci currently works around several readline/editline bugs.
• the global variable saved_sess will no longer be necessary due to monadic completion functions.

However, our dependencies on mtl, utf8-string, etc will have to be investigated.

[judah]

After some discussion on the mailing list, the following plan was proposed:

http://www.haskell.org/pipermail/glasgow-haskell-users/2008-November/016227.html

To summarize: for ghc-6.12 we plan to switch to haskeline, and also to implement a standard Unicode library which will replace the utf8-string dependency.

I'm closing this since there's now a task out on GHC's trac:

http://hackage.haskell.org/trac/ghc/ticket/2812