Prohibit unbuildable Windows configurations due to GHC bugs wrt TemplateHaskell and C interaction

[Bodigrim]

Closes #385.
@RyanGlScott, could you please take a look?

[RyanGlScott]

I'm no longer able to construct a build plan with GHC 8.2.2, so as far as I'm concerned, this fixes the issue for me. Thanks!

[Bodigrim]

@Boarders @Lysxia this is ready for review.

[RyanGlScott]

Is it possible to revise text-2.0 on Hackage to include these changes? Even after the release of text-2.0.1 (which includes this fix), cabal will still include text-2.0 in build plans involving GHC 8.2 on Windows, which breaks the build. See here for an example.

[Bodigrim]

I think Hackage revisions cannot add conditionals, so the only option is to deprecate text-2.0 altogether, which I do not feel fully justified at this point.

[RyanGlScott]

I think Hackage revisions cannot add conditionals

Darn, that's really unfortunate. Is there a hackage-server issue tracking this somewhere?

In the meantime, I suppose I'll just have to exclude GHC 8.2 from CI on Windows.

[phadej]

I think Hackage revisions cannot add conditionals

Darn, that's really unfortunate. Is there a hackage-server issue tracking this somewhere?

I cannot find one, but i'm pretty sure there was sometehing either on hackage-server or hackage-cli issue trackers. I was looking into that 2-3 years ago (it's not trivial, but not impossible either).

(relatedly, if hackage-trustees/hackage-cli#9 lib is split out into standalone lib, so it's actually shared between hackage-server and hackage-cli, might increase people motivation to contribute things. AFAIK those are also the only revision related tests existing - which is not that great)