@kacir - trying to clean up the "kitchen sink" an author at a time in the same library was also my initial approach. I can't recall all the things I experimented in terms of a tag or a custom column to make a book as processed but I also used a user category to mark an author as processed amongst them.
However the issue I found was twofold. Primarily that the Calibre performance was pretty dire - taking many seconds or sometimes minutes to perform simple operations. A lot of work has been done on Calibre performance since that time to improve things, but it made for intolerably slow performance at the time. Also back then I had not yet written a number of the plugins like Find Duplicates to assist.
The second reason was more the sense of lack of progress, every time I opened the library I was faced with the bulk of it being unprocessed, and the books I had done kind of got lost in the morass. Putting just the cleaned ones into a new library gave me a sense of achievement when I opened that up to know that all the books contained within were ready for reading. And also importantly the books I most wanted to be reading, rather than including many others that I had just collected for the sake of collecting. Of course you can use a search restriction to similar effect, but with the performance issues I had I chose a clean start instead.
As for the formats thing, I don't disagree that space is cheap but the more files you have you will pay a penalty somewhere. Whether it is at the time you do a search in Calibre, the time you do a search via the file system or doing a backup. As for the formats themselves, well it depends on whether you own a legacy reader (I don't) or not. Formats like lit, lrf, etc are dead as a dodo. Txt, rtf, doc are a poor man's ePub. HTML is a hop skip to an ePub. PRC, AZW - just use MOBI (which in itself is easily converted to/from ePub). These are my opinions based on the devices I own. Of course there will likely be some other format come out one day to obsolete ePub - but I am hedging my bets that given its dominance a product like Calibre will ensure you can convert from ePub to it. And quite apart from any of that - keeping say an ePub and a LIT format of a book will always leave a question in your mind as to which version is the "better" one if they were not converted from the same source. Which you can only resolve by inspecting. Better imho to do that job once, make your decision (converting the LIT to overwrite the ePub if it wins) and move on.
But we all have our own ways, that is part of the fun
