fix outlier removal bug (lightkurve#1313)

* fix outlier removal bug

* update test

* add nan to test

* update comment

---------

Co-authored-by: Christina Hedges <14965634+christinahedges@users.noreply.github.com>

src/lightkurve/lightcurve.py

@@ -1378,6 +1378,11 @@ def remove_outliers(
         # a local import here.
         from astropy.stats.sigma_clipping import sigma_clip
 
+        # astropy.stats.sigma_clip won't work with masked ndarrays so we convert to regular arrays
+        flux = self.flux.copy()
+        if isinstance(flux, Masked):
+            flux = flux.filled(np.nan)
+
         # First, we create the outlier mask using AstroPy's sigma_clip function
         with warnings.catch_warnings():  # Ignore warnings due to NaNs or Infs
             warnings.simplefilter("ignore")

tests/test_lightcurve.py

@@ -1009,7 +1009,6 @@ def test_remove_nans():
     lc_clean = lc.remove_nans("flux_err")
     assert_array_equal(lc_clean.flux, [])
 
-
 def test_remove_outliers():
     # Does `remove_outliers()` remove outliers?
     lc = LightCurve(time=[1, 2, 3, 4], flux=[1, 1, 1000, 1])
@@ -1025,6 +1024,10 @@ def test_remove_outliers():
     lc_clean = lc.remove_outliers(sigma_lower=float("inf"), sigma_upper=1)
     assert_array_equal(lc_clean.time.value, [1, 3, 4, 5])
     assert_array_equal(lc_clean.flux, [1, 1, -1000, 1])
+    # Ensure that we can sigma clip masked arrays
+    lc = LightCurve(time=[1, 2, 3, 4, 5], flux=Masked([1, 1, 1000, 1, np.nan]))
+    lc_clean = lc.remove_outliers(sigma=1)
+    assert_array_equal(lc_clean.time.value, [1, 2, 4])
 
 
 @pytest.mark.remote_data