Scrivener is designed for working in small (or small-ish) chunks. To use its linking/keyword/other functionality for the task you have in mind, you do need to divide your manuscript into smaller pieces, as already described. However, if you really don’t want to do that, there are more manual ways of achieving the same end result, and Scrivener has features that can help. Note that this suggestion has nothing to do with compiling topics with reference to their origin in the text (which can’t be done without implementing methods such as those suggested by Ioa and Katherine, and which all involve splitting your document). Instead, I am assuming that your ultimate objective is just to ensure that your new chapters haven’t omitted anything important from your first draft (in its single, long document), and that you don’t really mind how you achieve this as long as it gets done.
So, with those provisos, my suggestion is that you simply collate everything you have said in both your original draft and your new chapters about each character/topic, then compare the old and new material manually.
For example, create a new document for each of your characters/topics, so that you have documents called “music”, “Gus”, “Georgie” and “Amelia” (to use the examples you have mentioned). Go through your original text (make a snapshot of it before you start, in case you change something accidentally). Every time you find something about Gus that you want to make sure you have reflected in your new material, select that text then choose Edit > Append Selection to Document > Gus. Likewise for music, Georgie and Amelia, making sure that you are appending to the correct document each time. When you have finished going through the draft, you will have four separate documents containing everything important that you have said about each character/topic, in the order in which the details appear in your draft.
The next stage is to create similar documents for your new material, appending to new documents called “music new”, “Gus new” etc. When you have finished going through the new chapters in exactly the same way, this time appending to the " xxx new" documents, take a snapshot of each character/topic file (old and new).
Open “Gus” in one split window, and “Gus new” in the other, and analyse them in detail, seeing if each statement in the original draft is matched by one (or more) in the revised draft. Depending on the number of descriptions you have appended to each document, this could take a while, and it is very definitely a manual process (although Scrivener’s find and search functionality should be helpful). If you identify things from the old draft that have been handled in the new material, delete the corresponding statements from both “Gus” and “Gus new” (make sure that you catch all such occurrences in both documents, and don’t just stop at the first one – I always find that I repeat myself more than I realise, and leaving duplicates in place will only confuse things).
At the end of the process, anything that is left in “Gus” was in your original draft but does not appear in your new chapters, giving you a clear “to do” list for transferring missing information. Anything left in “Gus new” is in your new chapters but not in your original draft. And you still have your snapshots to refer to if you want to see everything that you picked out as important – each new addition is appended to the end of the document, so the order will give a degree of indication of position in the text.
This process is a bit laborious, but if you really don’t want to split your original document it might help you to achieve the end-result you want. It’s not really repeatable, without a lot of effort, but if the aim is just to make sure that the important stuff from your first draft is identified and incorporated into your revision, then you shouldn’t have to repeat the exercise again anyway. The first pass through the material will identify what you still need to incorporate, and you can check off those items as you work them into your revised draft.