I know what you're saying, but I did a quick test loading 1.2 million rows to a test table which is about 10x what the production table should ever be based on usage and maintenance. It went from sub second to under 2 seconds, and that with unfavorably structured data. Not ideal, but there are some other optimizations that could be made.

And I will say that Gail's approach would be fine. If it fit. It does not allow for matching on an unknown number of values per field.

And you'll have to come saw my foot off to get me to write dynamic SQL. 😉