I think you have contradicted yourself in your answer. First you say that you have an author name good enough that you can import it into calibre. Then you say that you cant use windows search because the author name doesn't always exist. Now unless you plan on relying on some metadata in the file when you bulk add to calibre (which you can never rely on imho, and it is only present in some formats etc) then you have the same "problem" either way?
If your input directories contain completely meaningless filenames, then your options aren't great regardless.
I make sure that as I get new books they always have a name I can later search on. Preferably in every filename, but worst case on the parent folder. Windows search can find either situation (unlike adding to calibre), it just means you have to drill into the author name folders to scrape up any books that didn't match the search.
Yep it is not a quick process. But unless you plan on buying every book all over again and putting in calibre immediately then I don't see any alternative. Had I been disciplined in rationalising my books as I accumulated them over the years as some people are, I wouldn't have an issue. Instead it was a "sort it out one day" job. Well one day has arrived. Hence why as I posted in your other thread ages ago I am just taking an author by author approach, and most days come home to chip away at a few more. I'm in no major hurry - I've got enough processed books to read for at least the next 10 years already
There is no magic bullet to make bulk importing happen by magic whereby calibre gives you this perfect metadata duplicate free best possible format version of a pile of kak input. We would all love there to be such a thing, but it doesn't exist. Garbage in, garbage out.