Update calibration.ipynb

selfcal -> self_cal when describing the function

docs/calibration.ipynb

@@ -109,7 +109,7 @@
         "id": "hdav__9fO17C"
       },
       "source": [
-        "## Run selfcal\n",
+        "## Run self_cal\n",
         "\n",
         "In this example, we are self-calibrating simulated full-polarization point-source visibility data.  The putative observation (based on a real ALMA dataset) had 47 antennas in one spectral window, with four correlations in the circular basis.  The data are corrupted by a systematic set of scalar multiplicative gain errors that are constant in time, frequency, and polarization.  Gaussian noise at a typical scale (appropriate for the recorded bandwidth and integration time, and an SEFD of \\~8400 Jy).  This example is perhaps atypical for self-cal in the sense that the data are not nominally calibrated to any degree, but since we can asumme a point source model, we can expect a good result.   In this case, we are assuming a 3 Jy (total intensity) point source model, which was stored with the imported data.  Since the gain errors affect all correlations in a systematic way, we can expect the calibration derived from the parallel hands to also properly calibrate the cross-hands, and thereby reveal the linear polarization of the point source (which we \"suspect\" to be 13%...)."
       ]