Use thumbnails, not the printed page numbers, to identify what you need.
Comma-separated: 1-3,7,9-12. Pages appear in the order you write them.
Check the page count and the first page of the new file before deleting anything.
Splitting every page gives you N files, one page each. Extracting a range gives you one new file containing only those pages. If what you want is a shorter document, extract the range — splitting first and merging afterwards works but wastes a step and loses the original page numbers.
The 1/50 in a reader's toolbar counts file pages. A document that prints "page 3" may sit at physical page 5 once the cover and contents are counted, so locate by thumbnail rather than by the number on the paper, or you will extract the wrong range.
A range like 1-3,7,9-12 is read left to right, so the output order matches what you typed. Ask for 9-12,1-3 and those pages come out in that order. That is useful for reordering, and a common cause of surprise when someone expects the tool to sort ascending.
Pulling ten pages out of a scan gives you those ten pages at the original resolution — pages are moved, not re-rendered. The file shrinks roughly in proportion to the page count, and anything smaller needs a separate compression pass. The reverse also holds: a blurry source stays blurry, so do not expect this step to rescue it.
No. The output is a new file and the source is untouched.
Yes. Extract everything except it, for example 1-4,6-20.
Not while it is encrypted. Decrypt it first.