Revert r45021, r45022, r45023 -- restoring r44896, r44927, r44896 -- pending further...

Revert r45021, r45022, r45023 -- restoring r44896, r44927, r44896 -- pending further discussion of what, if anything, is wrong with our ID generation. Given that the section IDs are meant to be valid IDs (as required per spec), and that in production they are already put out as IDs due to Tidy's behavior, the only change necessary to provide the fix (if fix is necessary in fact) would be to fix our ID normalization.