chore: update gimli

[philipc]

There were a few breaking changes in the gimli 0.33.0 release, and I've done this update to see what impact they had here. Most of these changes are straightforward. but a couple are notable.

UnitSectionOffset is no longer an enum. I've replaced all of its uses with DebugInfoOffset, since the code already didn't handle .debug_types, and it's unlikely it will ever need to.

next_dfs no longer returns the change in depth. The depth is now available from DebuggingInformationEntry::depth. However, it looks like the previous code was wrong: it treated the returned value as a depth, not a change in depth. So there may be a noticeable change in behaviour here which you may want to verify works as expected.

My own testing has been simply to check that CI passes.

[swananan]

Thank you very much for this PR, and sorry for the delayed response — I missed the notification and only noticed it when I came back to open another PR 😁. I've now started watching the repository so I don't miss something like this again.

I definitely didn’t expect to see a gimli update PR from you in GhostScope — that was both a surprise and very encouraging.

I especially appreciate you pointing out my incorrect use of depth. That was a very important catch, and it likely explains a case I had seen before where the relevant information was present in dwarfdump, but GhostScope did not pick it up. My current end-to-end tests clearly do not cover this scenario yet.

I'll merge this first, then follow up on my side with targeted tests and a careful pass through the surrounding logic, since I want to straighten out this whole path rather than just patch the immediate case. Thanks as well for flagging the gimli upgrade and for taking the time to send this PR — I really appreciate it.