Refactor batch sample matching

[benjeffery]

Refactor to work like ancestor matching.
Some feedback on the approach would be good here.

[codecov]

Codecov Report

Attention: Patch coverage is 91.50943% with 9 lines in your changes missing coverage. Please review.

Project coverage is 93.16%. Comparing base (1b88b5e) to head (abcab58).
Report is 4 commits behind head on main.

Files with missing lines	Patch %	Lines
tsinfer/inference.py	91.50%	5 Missing and 4 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #954      +/-   ##
==========================================
- Coverage   93.23%   93.16%   -0.07%     
==========================================
  Files          18       18              
  Lines        6299     6337      +38     
  Branches     1139     1133       -6     
==========================================
+ Hits         5873     5904      +31     
- Misses        290      294       +4     
- Partials      136      139       +3     

Flag	Coverage Δ
C	93.16% <91.50%> (-0.07%)	⬇️
python	95.53% <91.50%> (-0.12%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[jeromekelleher]

Looks good. I think some classes to capture the parameters would be helpful here, as the "metadata" dictionaries are confusing and easy to make mistakes with.

You could even add some functionality to the class to abstract out some of the common boilerplate, like

class WorkDescriptor:

     def common_params(self) -> dict:
            return {"precision": self.precision, ... etc}

match_samples(..., **wd.common_params())

[jeromekelleher]

I think it's worth making a dataclass to capture this now - it's getting a bit of out hand. "Metadata" is confusing as a name as I thought we were talking about "tskit metadata" initially. The thing we're talking about is a serialisable description of some work that's to be done... So,

@dataclasses.dataclass
class SampleBatchWorkDescriptor:
     sample_data_path: str
    ....

[benjeffery]

Fixed in 6924929

[jeromekelleher]

Looks good, happy to merge