Census -k incorrectly counts non-kern spines as notes

[craigsapp]

The "census -k" command incorrectly counts non-kern spines as notes. Consider the following data:

**kern	**dynam
1c	f
*-	*-

census -k on this input gives:

 HUMDRUM DATA

 Number of data tokens:     2
 Number of null tokens:     0
 Number of multiple-stops:  0
 Number of data records:    1
 Number of comments:        0
 Number of interpretations: 2
 Number of records:         3

 KERN DATA

 Number of note-heads:      2
 Number of notes:           2
 Longest note:              1
 Shortest note:             1
 Highest note:              f
 Lowest note:               c
 Number of rests:           0
 Maximum number of voices:  2
 Number of single barlines: 0
 Number of double barlines: 0

Notice that the census -k output says there are two notes when there is only one. Running the command:

 extract -f 1 | census -k

on the input data gives the correct count:

 HUMDRUM DATA

 Number of data tokens:     1
 Number of null tokens:     0
 Number of multiple-stops:  0
 Number of data records:    1
 Number of comments:        0
 Number of interpretations: 2
 Number of records:         3

 KERN DATA

 Number of note-heads:      1
 Number of notes:           1
 Longest note:              1
 Shortest note:             1
 Highest note:              c
 Lowest note:               c
 Number of rests:           0
 Maximum number of voices:  1
 Number of single barlines: 0
 Number of double barlines: 0

The census command should not count data in non-kern spines when the -k option is given. But pre-filtering the data with extract can work around the problem.