feat: support treetime ambiguous date format

Add support for the treetime ambiguous date format `[float_min:float_max]`, e.g. `[2004.43:2005.58]`
This is useful for when ambiguity is not limited
to month/year boundaries.
Resolves #1304

(cherry picked from commit df94593)

augur/dates/__init__.py

@@ -121,6 +121,12 @@ def get_numerical_date_from_value(value, fmt=None, min_max_year=None):
         except InvalidDate as error:
             raise AugurError(str(error)) from error
         return [treetime.utils.numeric_date(d) for d in ambig_date]
+    if value.startswith('[') and value.endswith(']') and len(value[1:-1].split(':'))==2:
+        # Treetime ambiguous date format, e.g. [2019.5:2020.5]
+        try:
+            return [float(x) for x in value[1:-1].split(':')]
+        except ValueError as error:
+            raise AugurError(str(error)) from error
     try:
         return treetime.utils.numeric_date(datetime.datetime.strptime(value, fmt))
     except ValueError:

tests/dates/test_dates.py

@@ -99,3 +99,7 @@ def test_get_numerical_dates_dict_error(self):
         }
         with pytest.raises(AugurError):
             dates.get_numerical_dates(metadata)
+
+    def test_get_numerical_date_from_value_treetime_ambiguous_date_format(self):
+        assert dates.get_numerical_date_from_value("[2019.5:2020.5]") == [2019.5, 2020.5]
+        assert dates.get_numerical_date_from_value("[2013.543:2013.544]") == [2013.543, 2013.544]