Fixed issue with score normalization with dev/eval groups

The current score normalization pipeline had an issue while executing experiments with dev/eval groups. Once the normalization statistics were computed on the dev set, its check-point contains a name that is indistinguishable from the one of the evaluation set. Consequently, the evaluation set pipeline tries to use the statistics from the dev set, which breaks the code.

This bug was introduced while the score normalization pipeline was refactored. It would be good in the future if we have a good dummy dataset to test all this stuff, even if this cost more hours on the CI.

bob/bio/base/pipelines/entry_points.py

@@ -337,15 +337,18 @@ def execute_pipeline_score_norm(
         )
 
     if checkpoint and not is_biopipeline_checkpointed(post_processor):
-        model_path = os.path.join(
+        score_stats_path = os.path.join(
             checkpoint_dir,
             f"{score_normalization_type}-scores",
             "norm",
-            "stats.pkl",
+            "stats",
         )
         # we cannot checkpoint "features" because sample.keys are not unique.
         post_processor = wrap(
-            ["checkpoint"], post_processor, model_path=model_path, force=force
+            ["checkpoint"],
+            post_processor,
+            model_path=score_stats_path,
+            force=force,
         )
 
     pipeline = PipelineScoreNorm(pipeline, post_processor)
@@ -355,6 +358,10 @@ def execute_pipeline_score_norm(
     # treferences = database.treferences(proportion=ztnorm_cohort_proportion)
     for group in groups:
 
+        # Changing the score normalization stats file name as a function of the group
+        if checkpoint and not is_biopipeline_checkpointed(post_processor):
+            post_processor.model_path = f"{score_stats_path}_{group}.pkl"
+
         if score_normalization_type == "znorm":
             score_normalization_samples = database.zprobes(group=group)
         elif score_normalization_type == "tnorm":