[dv] Fix paths in merge_cov.py

[superpung]

1. In metadata.py, class RegressionMetadata has a field named dir_cov, not cov_dir:
   

   ibex/dv/uvm/core_ibex/scripts/metadata.py

   Line 117 in 38da151

   dir_cov : pathlib.Path = field(init=False, compare=False, default_factory=pathlib.Path)

   
   

   ibex/dv/uvm/core_ibex/scripts/metadata.py

   Line 153 in 38da151

   self.dir_cov = self.dir_run/'coverage'

   
   But in merge_cov.py, it refers to a field named cov_dir that does not exist in class RegressionMetadata:
   

   ibex/dv/uvm/core_ibex/scripts/merge_cov.py

   Lines 45 to 47 in 38da151

   	'-dbname', str(md.cov_dir/'merged.vdb'),
   	'-report', str(md.cov_dir/'report'),
   	'-log', str(md.cov_dir/'merge.log'),

   
   This results in an AttributeError during dv process:

AttributeError: 'RegressionMetadata' object has no attribute 'cov_dir'

---

2. In scripts_lib.py, type of parameter cmd in function run_one is List[str]:
   

   ibex/dv/uvm/core_ibex/scripts/scripts_lib.py

   Lines 25 to 26 in 38da151

   	def run_one(verbose: bool,
   	cmd: List[str],

   
   But in merge_cov.py, it passes a list (list(cov_dirs)) with type List[pathlib3x.pathlib3x.PosixPath] to function run_one as a parameter:
   

   ibex/dv/uvm/core_ibex/scripts/merge_cov.py

   Lines 41 to 50 in 38da151

   	def merge_cov_vcs(md: RegressionMetadata, cov_dirs: Set[pathlib.Path]) -> int:
   	logging.info("Generating merged coverage directory")
   	cmd = (['urg', '-full64',
   	'-format', 'both',
   	'-dbname', str(md.cov_dir/'merged.vdb'),
   	'-report', str(md.cov_dir/'report'),
   	'-log', str(md.cov_dir/'merge.log'),
   	'-dir'] +
   	list(cov_dirs))
   	return run_one(md.verbose, cmd, redirect_stdstreams='/dev/null')

   
   This results in a TypeError during dv process:

TypeError: type of argument "cmd"[11] must be str; got pathlib3x.pathlib3x.PosixPath instead

[rswarbrick]

Hi @hcallahan-lowrisc! Would you mind taking a look at this? A quick glance at the code in question makes me think that the proposed fix is right, but I think your name is all over the code in question :-)

[hcallahan-lowrisc]

Ah okay yes this appears to be exercising some code that we don't run here regularly. We use Xcelium for our regressions, so it appears the vcs flow has not been tested.

I'll test this locally and get back to you, but the fix looks reasonable to me.

[hcallahan-lowrisc]

Great, thanks for the PR!