Test helper: Mark current run as reference

When developing on edalize, it's very common to make changes which
affect the generated files. To update the tests, one then needs to
manually apply the changes to the generated files, or step-by-step copy
over the generated tests from the failing comparison to the reference
location.

This commit adds a way to automate this process by marking a pytest run
as "reference run", and using the generated files as reference file.

To mark a run as reference run, set the GOLDEN_RUN environment variable
to any value, e.g.

```
GOLDEN_RUN=1 pytest
```

Don't forget to use git diff to then compare if the changes are actually
expected!

Implementation note:
This uses an environment variable and not a command line argument for
ease of implementation. Command line arguments needs to be passed as
fixture in pytest, which requires changes to every test_* function we
have, or ugly workarounds with globals.

tests/edalize_common.py

@@ -7,16 +7,21 @@
 
 def compare_files(ref_dir, work_root, files):
     import os.path
+    import shutil
 
     for f in files:
         reference_file = os.path.join(ref_dir, f)
         generated_file = os.path.join(work_root, f)
 
         assert os.path.exists(generated_file)
 
+        if 'GOLDEN_RUN' in os.environ:
+            shutil.copy(generated_file, reference_file)
+
         with open(reference_file) as fref, open(generated_file) as fgen:
             assert fref.read() == fgen.read(), f
 
+
 def param_gen(paramtypes):
     args = []
     defs = {}