Why not just UNION all your tables into one datasource?
The tables are of the same structure (at least for this analysis),
so there would be just one [date] column in the final datasource,
and your COUNTD() aggregations (and their Ratio calcs) would work seamlessly.
And you'd be able to distinguish the original tables by the [Table Name] column.
To UNION tables from different Connections (you've got Excel and BQ),
you may want to apply the 'Scaffold' technique by Jonathan Drummey: