Time domain data

Update time domain data processing so that the distribution of log noise evidence follows the expected distribution.

This is mostly formalising where and how things are windowed and fixes some broken internal references.

I also bumped into this issue, https://git.ligo.org/Monash/tupak/issues/133.

test/detector_tests.py

@@ -373,7 +373,7 @@ class TestInterferometerStrainData(unittest.TestCase):
             time_domain_strain, sampling_frequency)
 
         self.assertTrue(np.all(
-            frequency_domain_strain == self.ifosd.frequency_domain_strain))
+            self.ifosd.frequency_domain_strain == frequency_domain_strain * self.ifosd.frequency_mask))
 
     #def test_sampling_duration_init(self):
     #    self.assertIsNone(self.ifo.duration)