Refactoring quantity index table by removing the canonical columns

[jingtang10]

I don't mind keeping only canonicalized quantities in Index. Could we make it explicit about not supporting case 4 in @epicadk comment?

Another possible option would be to not support canonical match at all as discussed with @jingtang10 . I think that makes more sense than this.

My guess is that it's unlikely for users to search by a quantity using the equal operator. I would imagine quantity would be used for sorting. But also probably used for search within a range.

If this is true, I would argue for canonicalisation. Although to make sorting work perfectly with different units there will be much more work.

Originally posted by @jingtang10 in #682 (comment)