First, KDE gets their features merged into Qt and released in a matter of months also.
Second, KDE doesn't have to be in control of Qt to get things done. Qt is easily extendible. If KDE wants a new widget, they can just make it. If it isn't general purpose enough to be incorporated into Qt or is for some reason rejected, KDE can provide it on their own. KDE has always maintained its own special-purpose widgets and always will. And KDE is not the only group doing this, there are a lot of groups providing special-purpose add-on widgets for Qt (Qwt, for example), and in fact there is a community developing around such add-ons for Qt.
This is a limitation of GTK, not an advantage. One of the complaints you hear over and over from groups switching from GTK to Qt is that it is extremely difficult, convoluted, and poorly-documented for outsiders to make new or significantly modified widgets for GTK, while it is trivial in Qt. Whether the difficult in supporting custom widgets has forced Gnome and GTK closer together or whether Gnome and GTK being close has made them unconcerned with supporting custom widgets, the end result is that if you want something that isn't in GTK, your only viable option is to convince Gnome to implement it for you. For Qt, that is an option, but not a hard requirement.