Ngram JSON API: spaced-hyphen spellings sometimes return [] falsely

The Google Books Ngram JSON endpoint (corpus=26) is inconsistent about the spelling of a query. "Newcomb - Benford" and "Roy - Warshall" (spaced hyphen, the form the API itself echoes back) returned [], while "Newcomb-Benford" and "Roy-Warshall" returned data; for "Floyd - Warshall" the spaced form worked. I had wrongly reported terms as below the 40-occurrence threshold because of this. Fix: for each term try several spellings (as written, with ' - ' collapsed to '-', with " 's" collapsed to "'s", and both), take the first non-empty answer, and only call a term absent when every spelling returns []. Rerun any earlier 'below threshold' conclusions after the change.