Cool, so we're in agreement that just because X unit might not be comfortable with Y group, that's insufficient reason for removing Y group from the ability to give active service. Don't ask, don't tell was removed because it was unconstitutional, no matter who had problems with homophobia within the ranks.
Don't forget exactly the same arguments were made about desegregating the military back after the Second World War.
I was just out of the army when Australia decided to allow women into combat roles. I was against it, sincerely believed the "but unit effectiveness!" rhetoric. It's been several years now, women are integrated in Australian combat units, and all the doomsayers (like me) turned out to be just as wrong as the doomsayers who said letting women into the Army ion the first place would be bad, or doomsayers saying removing racial segregation would be bad before that. So, trying to learn from earlier mistakes, I'm open to "unit effectiveness" arguments about transgender soldiers, but I'm going to want to see some pretty thorough data before buying the argument, rather than the old "but everybody knows!" and "well some troops just plain don't like 'em!" rationalisations.
Professional soldiers are grown ups. You don't have to LIKE who you're serving with, that's what discipline is for.