ENH: Properly handle NEXRAD varying moments

The moments vary from sweep to sweep, but we were always reading 6
pointers rather than looking at the encoded value. This would have
bitten us when they started shipping a 7th moment with Build 19.

src/metpy/io/nexrad.py

@@ -581,9 +581,10 @@ def _decode_msg31(self, msg_hdr):
         msg_start = self._buffer.set_mark()
         data_hdr = self._buffer.read_struct(self.msg31_data_hdr_fmt)
 
-        # Read all the data block pointers separately. This simplifies just
-        # iterating over them
-        ptrs = self._buffer.read_binary(6, '>L')
+        # Read all the data block pointers separately. This makes it easy to loop and to
+        # handle the arbitrary numbers. We subtract 3 for the VOL, ELV, and RAD blocks that
+        # are required to be present (and can't be read like the data)
+        ptrs = self._buffer.read_binary(data_hdr.num_data_blks - 3, '>L')
 
         if data_hdr.compression:
             log.warning('Compressed message 31 not supported!')

tests/io/test_nexrad.py

@@ -23,22 +23,24 @@
 # 1999 file tests old message 1
 # KFTG tests bzip compression and newer format for a part of message 31
 # KTLX 2015 has missing segments for message 18, which was causing exception
-level2_files = [('KTLX20130520_201643_V06.gz', datetime(2013, 5, 20, 20, 16, 46), 17),
-                ('KTLX19990503_235621.gz', datetime(1999, 5, 3, 23, 56, 21), 16),
-                ('Level2_KFTG_20150430_1419.ar2v', datetime(2015, 4, 30, 14, 19, 11), 12),
-                ('KTLX20150530_000802_V06.bz2', datetime(2015, 5, 30, 0, 8, 3), 14),
-                ('KICX_20170712_1458', datetime(2017, 7, 12, 14, 58, 5), 14),
-                ('TDAL20191021021543V08.raw.gz', datetime(2019, 10, 21, 2, 15, 43), 10)]
+level2_files = [('KTLX20130520_201643_V06.gz', datetime(2013, 5, 20, 20, 16, 46), 17, 4, 6),
+                ('KTLX19990503_235621.gz', datetime(1999, 5, 3, 23, 56, 21), 16, 1, 3),
+                ('Level2_KFTG_20150430_1419.ar2v', datetime(2015, 4, 30, 14, 19, 11), 12, 4, 6),
+                ('KTLX20150530_000802_V06.bz2', datetime(2015, 5, 30, 0, 8, 3), 14, 4, 6),
+                ('KICX_20170712_1458', datetime(2017, 7, 12, 14, 58, 5), 14, 4, 6),
+                ('TDAL20191021021543V08.raw.gz', datetime(2019, 10, 21, 2, 15, 43), 10, 1, 3)]
 
 
 # ids here fixes how things are presented in pycharm
-@pytest.mark.parametrize('fname, voltime, num_sweeps', level2_files,
+@pytest.mark.parametrize('fname, voltime, num_sweeps, mom_first, mom_last', level2_files,
                          ids=[i[0].replace('.', '_') for i in level2_files])
-def test_level2(fname, voltime, num_sweeps):
+def test_level2(fname, voltime, num_sweeps, mom_first, mom_last):
     """Test reading NEXRAD level 2 files from the filename."""
     f = Level2File(get_test_data(fname, as_file_obj=False))
     assert f.dt == voltime
     assert len(f.sweeps) == num_sweeps
+    assert len(f.sweeps[0][0][-1]) == mom_first
+    assert len(f.sweeps[-1][0][-1]) == mom_last
 
 
 def test_level2_fobj():