Well, not a particularly sophisticated idea but you could try and use a local H2 database to join two datasets by time ranges. Depending on the power of your machine and other factors that might work. I tried it with files of different sizes and at least it gives you a result.
Until there is a solution in KNIME you could explore further the possibilities of using SQL to make the join more efficient. I took from this older entry:
There might be some special limitations when working with a local H2 database. Depending on the power of your machine you could try and load the H2 DB into memory which might speedup the process.