CSS class identifiers should be cleaned prior to be applied to regions

Repeatability: always

Steps to repeat:

1. Find a token which resolves to two separate words. Say for instance that the [node:status] token resolves to Not published (even though it can actually resolve to unpublished -- this is just an example. One could use any other token that yields characters not normally used for CSS identifiers.)
2. In Display Suite, create a CSS class using that token, for example ds-node-status-[node:status].
3. In the Manage Display section of a content type, assign this class to a region.
4. View the content, inspect the markup.

Expected results:
The region is assigned a single class ds-node-status-Not-published.

Actual results:
The region is assigned two separate classes, such as ds-node-status-Not and published, because of the space character in the string returned by token_replace. I realize I can then use the CSS selector div.ds-node-status-Not.published, but I don't think this is the intended way it's supposed to work.

Proposed (vague) solution:
The Not published string returned by token_replace should be sanitized/cleaned, perhaps using Drupal's drupal_clean_css_identifier function? I assume this should be used somewhere in the function ds_entity_variables of ds.module. I haven't been able to produce a working patch because I am a beginner in php/Drupal. I'm hoping someone more skilled will be able to propose a concrete solution.