Add test for symlink resolution

[furtib]

Test for #14

The problem was that paths in plist files didn't get resolved when there was an error in a header file.

To reproduce the bug, I had to:

• Add a header file that contains an error
• Add a source file that includes said header
• Add a rule for the new source fileí

Depends on: #60

[nettle]

I think we should not submit this until we agree on #22

[furtib]

Updated in accordance with #22, since this is a test for #14 it's expected to fail!

[Szelethus]

I don't think its wise to copy-paste so much from each test file to the next (grep_file(), TestBase etc), we should think on this a little more. Sorry for being such a bore, we'll figure this out and we'll happily ride onto the sunset :)

Otherwise, the main idea behind this solution (directory structure, the test itself) looks great. Its just the boring stuff.

edit: some ideas here: #22 (comment)

[furtib]

I tried your approach, about the common lib, should that be its own PR? Or should we just leave it here, and then every other test will depend on this one?

[furtib]

Depends on #61

[Szelethus]

Can you tidy this up with some python formatter? Otherwise, the direction is right.

[Szelethus]

Use some formatter to make the code uniform and <80 cols. Otherwise LGTM

[Szelethus]

Very well done @furtib. This was a lot of work, and I think it shows. Lets keep our eyes open if we need to adjust on the testing infrastructure, but for now, this looks pretty solid.