feat(observation): expose spell cast time in the snapshot

The per-spell Observation dict gains the additive cast_time key: the
registry spell's budget code on a hit, "standard" on a miss (mirroring
the engine's resolver fallback), so an LLM policy can schedule casts
against the action budget it sees reported back.

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

tests/test_spell_area_snapshot.py

@@ -127,7 +127,7 @@ def test_area_snapshot_existing_keys_unchanged() -> None:
     """Given: a known Fireball snapshot dict
     When: all keys and values are checked
     Then: the nine pre-U7a keys are present with unchanged values, and the
-    exact key set is the old nine plus the additive "area"."""
+    exact key set is the old nine plus the additive "area" and "cast_time"."""
     s = _spell_dict_for(_FIREBALL, "Fireball")
     assert s["name"] == "Fireball"
     assert s["level"] == 3
@@ -140,6 +140,7 @@ def test_area_snapshot_existing_keys_unchanged() -> None:
     assert s["summary"] == "area 20ft"
     assert s["half_on_save"] is True
     assert s["save_negates"] is False
+    assert s["cast_time"] == "standard"
     assert set(s.keys()) == {
         "name",
         "level",
@@ -151,6 +152,7 @@ def test_area_snapshot_existing_keys_unchanged() -> None:
         "half_on_save",
         "save_negates",
         "area",
+        "cast_time",
     }