It's a terrible client. The only reason it's survived so long despite the owner shamelessly milking it of every penny they can is name recognition. The brand isn't being tarnished - it's being sacrificed wholesale for dollars, with the singular goal of extracting as many dollars as possible before it dies.
I have been using qBittorrent (http://www.qbittorrent.org/) for a while and don't have anything to complain about.
Perfect for my simple needs (RSS feeds, launch a program on completion).
It does it's job well enough, ads can be easily disabled via advanced settings, it works in portable mode without even having a chance to install bundled crapware, the UI does not make my eyes bleed.
A philosophical difference - I believe that being forced to jump through hoops like that just to make it usable are sufficient to make it terrible. These sorts of things are designed to make it just OK enough to use for technically inclined users, while profiting off of those who don't know better - what percent of users do you think go through the steps you list?
You can also add hosts file entries to block application ads after dumping network traffic to find the servers, you can edit binaries with a hex editor and debugger to patch out code paths. But should you need to, just to make it usable? What amount of work is needed to make it not-terrible before it can be declared terrible?
Here's some off the top of my head. I know they're nit picking, but that's pretty much what you have to do to create a decent UX
* When you're under active torrents, select a torrent, wait for it to become unactive (ie going from downloading to done), the torrent is unselected (because it's not in view anymore) and the details of the torrent are not available. With µ if you select a torrent it's selected until you select something else no matter what states the torrent goes through and which view you're in.
* When adding a new torrent, the default destination is always the same, in µTorrent it's the last place you downloaded. When I add a torrent, the content always goes into a sub folder depending on type, it's much more likely that the last destination I used is the correct one (say I want to download a couple of music albums)
* The UI when adding a torrent is just cluttered and confusing and I have, a couple of times managed to enter the "add torrent-file"† instead of the "select destination for download" dialogue because it's such a mess
† Why is this even accessible from the prompt I get when I double click on a torrent in my file explorer? When would I ever double click on a torrent file, and then want to select a different torrent file? So unnecessarily messy.
I develop. Did not know Deluge was using Python. I might contribute one day, but doing front-end stuff is not really something I enjoy
I suggest Deluge, or maybe Transmission.
http://deluge-torrent.org/
edit: retracting Transmission recommendation because I'm unfamiliar with it's state on Windows. Looks like there are builds, but I've never used them.