The spaces at the end of segments have no significance, generally speaking. They can be either present or absent, although we might want to automate a “cleaning” process to ensure it is standard. The only tricky thing is the em-dash, which we have discussed. Otherwise a client can simply add the spaces automatically.
My own thought is that we should probably build a bilara-data auto-fixer-upper thingy, which would scan the repo and do things like correct quote marks, turn … into …, turn – into – and — into —, eliminate double spaces, and standardize trailing spaces. My preference would be that everything in bilara-data is correct as-is requiring minimum transformation by a client, so that would mean adding trailing spaces except where a segment ends in em-dash.