I don't mind the multiple stages in the Alps and Pyrenees. What I mind is that they go over the same parts of those ranges (especially the Pyrenees) year on year leaving vast swathes of them unused. A lack of variety means everybody knows the climbs like the backs of their hand, where to go, how to dose their efforts, so there isn't the element of surprise.
You could definitely do a stage that uses the big climbs early and then easier ones later on. A stage like
2009 Tarbes or
2010 Pau is simply not going to work in modern cycling unless you weaken the teams, especially as the latter came the day before a big MTF so the major contenders weren't interested. Simply as the grimpeurs are not going to be able to keep an organized péloton at bay for long enough to make it worthwhile, so you need some more climbs in the run-in.
The 'hard mountain-easy mountain' stratagem is a popular one in the Race Design Thread, inspired by the excellent racing produced by this method, whereby the final mountain is not particularly selective on its own, but the penultimate climb is particularly hard in order to encourage moves here, but also mean that if the group is together, tired legs make the final climb more decisive. The ultimate example is obviously Mortirolo-Aprica; Aprica on its own is not a tough climb - in fact the Giro Rosa had a one-climb Aprica stage as a "hilly" stage in 2015, with Pauline Ferrand-Prévot taking just a couple of seconds' gap over a chasing group of around 25 of the élites - and the climbing depth in the women's péloton is much more limited than that in the men's, as the kinds of gaps opened by the Mortirolo and Signora della Guardia stages this year showed, where the bunch was blown to smithereens 50km from the line (the latter was an AWESOME stage, by the way).
Andorra gives us the possibility to do something like this, by going with the brutal climbs like La Gallina or the normal tough climbs like La Rabassa, but instead of having a ski station MTF, you could finish after La Comella. I proposed
this stage in one of my Vuelta designs, but you could leave Beixalis out and have just La Comella if you liked, to create a different type of stage finish. The PRC guys cam up with
this stage - leave out Beret at the end and it fits your criteria perfectly. It also upset me because a while ago I was looking for a way to use Beret more interestingly than usual and the stage I came up with was almost identical to that only I didn't use Pla Batalher and I started in Bagnères-de-Luchon. My most recent Race Design Tour had a stage in the Vosges that was mainly about big mountains early, medium mountains late, and looked like
this although still with an uphill finish might be outside of your wishes, as I'd guess you'd be looking for something more like the Giro di Lombardia with some tougher climbs early on.