Remove MypyItem.mypy_path

[dmtucker]

Fix #58

First, I attempted to fix MypyItem.mypy_path; however, Mypy wasn't very helpful, unfortunately.
The problem is that the code it uses to calculate the path it ultimately displays is neither conveniently grouped nor exposed in the public API. I got a copy-pasta solution working (though, it is not fully covered) in 6464322, but concerns about the maintenance burden of trying to re-create what Mypy does under-the-hood (especially if that ever changes incompatibly) begged for another solution.

The primary reason for MypyItem.mypy_path is to match the path found in Mypy output to the appropriate MypyItem. It is also used in reportinfo, which is meant to reflect the path from Mypy output (e.g. foo in foo:5: error: from mypy).

The first need can be satisfied (with os.path.abspath) by normalizing the MypyItem and Mypy output paths before matching them. As for reportinfo, the point of the header, IMO, is to clearly correspond to results produced by Pytest during a run. For example:

========================= test session starts =========================
platform darwin -- Python 3.4.10, pytest-4.6.6, py-1.8.0, pluggy-0.13.0
Using --randomly-seed=1572200516
rootdir: /Users/davidmtucker/Projects/pytest-mypy
plugins: cov-2.5.1, randomly-2.1.1, mypy-0.4.1
collected 1 item                                                      

Running mypy on 1 files... done with status 1

src/pytest_mypy.py F                                            [100%]

============================== FAILURES ===============================
_________________________ src/pytest_mypy.py __________________________
5: error: No library stub file for module 'pytest'
5: note: (Stub files are from https://github.com/python/typeshed)
====================== 1 failed in 2.60 seconds =======================

Here we see an F in the output:

src/pytest_mypy.py F                                            [100%]

The header should clearly allow me to associate the subsequent report with that F. This is a very simple example, but as the test run grows larger, I believe this becomes even more important. Pytest associates that F with src/pytest_mypy.py (which seems to be calculated here: https://github.com/pytest-dev/pytest/blob/8ccc0177c8be85c0725981b3a9e867baeebfbe33/src/_pytest/terminal.py#L290). So, making that the header seems natural:

_________________________ src/pytest_mypy.py __________________________

Since MypyItem.mypy_path is technically public, this is a backwards-incompatible change, so it will target 0.5.0 instead of 0.4.2.