explore multi term matching in TermNormIsSubString

[Richard Jackson]

Currently this strategy only maps to a single term.

Is this what we want, or should we allow multiple-token substrings straight away?
Can we document this in the docstring and say why?