Generic and Dewey call number normalizers should sort volume numbers numerically, not asciibetically

[Galen Charlton]

Consider three volumes on the same bib with the following call numbers:

234.54 ABC vol. 1
234.54 ABC vol. 2
234.54 ABC vol. 10

Using the generic normalizer, the label_sortkeys would be

234.54 ABC VOL. 1
234.54 ABC VOL. 2
234.54 ABC VOL. 10

Using the Dewey normalizer, the label_sortkeys would be

234_540000000000000_ABC_VOL__1
234_540000000000000_ABC_VOL__2
234_540000000000000_ABC_VOL__10

In both cases, the effective sort order is:

234.54 ABC vol. 1
234.54 ABC vol. 10
234.54 ABC vol. 2

which is not ideal.

What I think the desired behavior is:

[1] In the case of the Dewey normalizer, groups of numbers that occur after the main classification number should be left-zero-padded.
[2] In the case of the generic normalizer, all numeric tokens should be left-zero-padded.

Evergreen master

[Dan Scott (denials)]

Or call number normalization routines could normalize call numbers
according to their classification definitions, as they're doing now,
and suffixes like "Vol. 10" that don't belong to the call number
proper could either be added as a call number suffix (now that we've
had that option for a year or so on a per-call number basis), or they
could be catalogued as parts (which we have also had for a year or
so).

[Galen Charlton (gmc)]

The LC normalizer already currently zero-pads numeric elements that occur after the classification part. For obvious reasons of history if nothing else, use of call number suffixes is not universal, and monograph parts are not used by serial units.

[Galen Charlton (gmc)]

Also, analytic call numbers can have volume designations that are considered part of the main call number, not enumeration designation that could be put in a suffix or hypothetical enumeration field.

[Ben Shum (bshum)]

I know our catalogers would appreciate not having to recreate everything multi-volume related into parts or use suffixes. So having the label_sortkey add padded 0's to help sort things out sounds pretty good to us.

[Chris Sharp (chrissharp123)]

Just noting here that we're still seeing this problem in 2.11.1.