feat: add Feature class for extracted AV features

This composes an Interval with some ancillary data:
 - source (what feature extractor did we get this from)
 - path (where is it now on the FS)
 - score (for ranking)

TODO double check we aren't overloading the word 'source' to have two
meanings

pipeline/utils.py

@@ -1,3 +1,5 @@
+import subprocess
+
 class SourceMedia():
     """Source media used by eg feature extractors. This is a list of Source objects.
 
@@ -114,6 +116,7 @@ class Interval():
 
     DEFAULT_DURATION = 5 # seconds
     DEFAUT_PRECISION = 3 # decimal places
+
     def __init__(self, start=None, end=None, duration=None):
         if start is None and end is None and duration is None:
             raise ValueError("Two of start, end, or duration must be provided")
@@ -196,3 +199,65 @@ class Interval():
             self.duration = new_duration
         self.end = self.start + self.duration
 
+class Feature():
+    """A feature extracted from a media file ("has a" Interval)
+
+    This extends intervals by adding other fields, such as the feature source and 'score'
+
+    Instance variables:
+
+    interval -- Interval: time of feature in the media file
+    source -- the source of the feature (ie feature extractor) (default: "unknown")
+    path -- the path to the media file
+    score -- the score of the feature (eg laughter confidence score, [0, 1] = { x ∈ ℝ | 0 ≤ x ≤ 1 }) (default: 0.0)
+
+    Notes:
+
+    - score is notionally in the closed interval [0, 1], but this is not enforced -- it is up to the feature extractor to ensure this (or use scores outside this range if desired -- eg a feature manually selected by user input might have a score of 2.0 so it is sorted 'above' other features)
+    - sorts based on interval, then source, then score
+    - path should never be unknown, since we need it to make clips from
+    """
+    # TODO: consider renaming score to something more generic
+
+    def __init__(self, interval=None, source=None, score=None, path=None):
+        """Create a feature with an interval, source, and score
+
+        Expects a ready-made interval; source and score are optional
+        """
+        if interval is None:
+            raise ValueError("Interval must be provided")
+        self.interval = interval
+
+        if path is None:
+            raise ValueError("Path must be provided")
+        self.path = path
+
+        if source is None:
+            source = "unknown"
+        self.source = source
+
+        if score is None:
+            score = 0.0
+        self.score = score
+
+    # classmethods for creating a feature with an interval directly
+    # which delegate to the Interval class :)
+
+    @classmethod
+    def from_start(cls, start=None, source=None, score=None, path=None):
+        return cls(interval=Interval.from_start(start), source=source, score=score, path=path)
+
+    @classmethod
+    def from_end(cls, end=None, source=None, score=None, path=None):
+        return cls(interval=Interval.from_end(end), source=source, score=score, path=path)
+
+    def __repr__(self):
+        return f"Feature({self.interval}, {self.source}, {self.score})"
+
+    def __lt__(self, other):
+        """Sort based on interval, then source, then score"""
+        if self.interval == other.interval:
+            if self.source == other.source:
+                return self.score < other.score
+            return self.source < other.source
+        return self.interval < other.interval