Ark 1 is supposed to sound unruly and aggressive, ive never felt the need to disable an RR from it. If you want smooth - buy berlin strings? The QA is a whole lot different considering they are dedicated libraries that cost more than an entire Ark on their own - and unlike Ark - it's designed to be versitile.
If Ark 1 was supposed to be the cheap entry level bundle with poor QA, it wasn't communicated well to me. There's aggressive, and then there's "I can't make a fast repetition line on this note without manually re-aligning one of the note-ons so that it doesn't sound like a random break was thrown in". They know there are errors like this, because another user from here reported this to them. They said they'd fix it in a future update, which hasn't come yet. And I doubt they're doing a proper QA pass looking through all the samples of the shorts this time either.
And I would argue that the stuff that bothers me in terms of other inconsistences sounds straight up bad at times, but I concede that there's some level of subjectivity involved. If you never wanted to disable a RR from it, you might be layering many things on top of each other, whereas I usually use it more exposed because I don't have that huge library collection and I bought it as an allround orchestral library for loud stuff.
The more practical thing if a particular sample bothers you - is to just disable it whenever it comes up in a phrase, render in place, then re-enable it.
Rendering to wav would mean I can't make changes to the microphone settings without manually re-doing all those individual tweaks, that's not an option for me.
It's probably possible to open up the instrument, delete the mapping group on those 2 semitones and extend the above and below ranges a half step a piece.
You're assuming that both neighboring notes don't contain RRs that bother me and that the sound of them being pitch shifted 2 semitones wouldn't bother me, and I'm not so sure about both. But thanks for the recommendations, I do appreciate them!
I think I might try the following, since most of the RRs that bother me are in the string spiccatos: I'll keep using my "articulation by midi channel" setup for my main work and throw in an RR reset trigger at the start of every phrase, and when somewhere a RR sample is hit in a spiccato part that I really can't stand, I'll move that part over to a midi track that has just high and low spiccatos and all RR layers activated on channel 1 and individual RR layers on channels 2 and upwards. Then I can maybe leave most of the part of the phrase on channel 1 and only "evade" individual notes that would trigger bad RRs.
P.s.: At some point in the future I'll probably buy the Cinematic Studio Strings/Brass/Woods. From what I've read that seems to be the best fit for me so far.