'stack ghci': build project before launching GHCi

[borsboom]

(closes #1065, fixes #1180)

Optimistically, launch GHCi anyway even if the build fails.

@chrisdone, can you review?

[borsboom]

@chrisdone @mgsloan: Is building the project (and continuing even if it fails) the correct behaviour for stack ide?

[mgsloan]

Actually, I think the right behavior for stack ide is to only build dependencies.

[borsboom]

@mgsloan: Should it stay with the old behaviour of not starting if the dependencies fail to build, or would the new behaviour of continuing anyway be better?

[mgsloan]

The new behavior of continuing is better, I think. At least that way the user ends up with a GHCI session that's somewhat setup.

[borsboom]

I've rebased this to resolve merge conflicts. @chrisdone / @mgsloan: still not sure what the right behaviour for stack ide and stack targets should be.

[mgsloan]

LGTM! should the --dependencies-only alias be part of the BuildSubset parser?

[mgsloan]

I think the check for (null realTargets) should be removed. See point 1 on #1249 for why. See also #817 for why the check was there in the first place.

[borsboom]

@mgsloan I've updates based on your comments, please give this a final review and merge if good.

[mgsloan]

LGTM!