Yep, resource conservation is key. It's a great thing to catalogue everything.
It's a shame that Sierra games have so many bugs. Also the original translations are very poor. I think the worst in terms of bugs is QFG4.
I have a question: When the version and the interpreter are the same, but the hash is different, is it the same game or is it an alternative version?
I grabbed that archive. Painfully slow going through archive.org. It will take sometime to go through the fluxes to add them into my collection. I have not done any new streams for a while since my 5.25" floppy drive died and the price on them has shot up.
As to the Version Tool, I am starting to go through it to see what releases it has that are missing in my collection. It did have one of the earliest versions of KQ4, version 1.000.106, interpreter 0.000.247. It predates SCI's ability to accept external patch files. There does seem to be one earlier release that has interpreter 0.000.247, but I don't know what game version it is or anything more about it. If anyone knows about it, please let me know.
Where are you getting the different hash? Through the tool? If so, the hash generator accessible through the tool menu item was simply generating the wrong hash. That is the problem with a project that has say on the back burner for years. You forget things. The SCI method to get the MD5 hash does what SVM does. It only reads up to the first 5000 bytes of the target file into the byte array. This is to speed things up when scanning larger files. It also generates hashes consistent with SVM. The stand alone hash generator was using the entire file. The current version, which I will probably upload towards the end of the week uses the same method.
ETA: the MPC version of QfG4 is far less buggy than the floppy version. It still has a couple of timer bugs that are solved by the NRS patches.