Allow Sessions to track volume ID instead of volume data

This is work towards #57
When loading a Session, I think it makes sense to allow the volume
data to remain a "reference" to a file rather than load the actual
data into memory. This way the users of the Session api can decide
how they then want to handle volume I/O.

Author: ebrahimebrahim

src/openlifu/db/session.py

@@ -24,7 +24,7 @@ class Session:
     date: Date of the session
     targets: sonication targets
     markers: registration markers
-    volume: volume
+    volume: loaded volume data
     transducer: transducer
     attrs: Dictionary of attributes
     date_modified: Date of last modification
@@ -35,7 +35,8 @@ class Session:
     date: datetime = datetime.now()
     targets: List[Point] = field(default_factory=list)
     markers: List[Point] = field(default_factory=list)
-    volume: xarray.DataArray = field(default_factory=xarray.DataArray)
+    volume: Optional[xarray.DataArray] = None
+    volume_id: Optional[str] = None
     transducer: Transducer = field(default_factory=Transducer)
     attrs: dict = field(default_factory=dict)
     date_modified: datetime = datetime.now()
@@ -81,8 +82,13 @@ def from_dict(d:Dict, db:'Database'):
             d['date'] = datetime.strptime(d['date'],DATE_FORMAT)
         if 'date_modified' in d:
             d['date_modified'] = datetime.strptime(d['date_modified'],DATE_FORMAT)
-        if isinstance(d['volume'], dict):
-            d['volume'] = xarray.DataArray.from_dict(d['volume'])
+        if 'volume' in d:
+            if isinstance(d['volume'], dict):
+                d['volume'] = xarray.DataArray.from_dict(d['volume'])
+            elif isinstance(d['volume'], str):
+                d['volume_id'] = d['volume']
+                # If we kept this key then it would assign a string to the volume attribute, which the wrong type:
+                d.pop('volume')
         if isinstance(d['transducer'], dict):
             transducer_id = d['transducer']['id']
             transducer  = Transducer.from_file(db.get_transducer_filename(transducer_id))