On the image width aspects alone I agree, althought I'd counter that mixing in the SBD mostly takes care of the middle and allows one to err to the wider side of things with the on-stage pair to improve other 'non-imaging' aspects.
What's a good configuration for the on stage mics? Can cards be used there or are omnis the better option?
It's a question of what you're recording and how its setup. For 2ch recording on stage, I've used cards and subcards before. You use a tighter stereo pattern since you're not getting similar information on both sides of the image. If I normally liked 20cm/90* then for onstage (if I could fit everything close to 160* for a stereo image) I'd condense that to 14cm/60* and still get a major stereo angle. Omnis don't matter nearly as much as the only directional stuff is >4khz and even that's sort of omni at the end of the day. The downside to omnis are that they don't have any rejection of those screaming fans in the front row.
This is one of those instances where if you want to get better at stereo onstage recording, I'd recommend learning the whole SRA calculation bit. Unless someone beats me to it, I'll dig it up later.
Do you points caps level or up?
At whatever I'm recording (not intending to be snarky). If that's above me, then up, if it's off to one side, then I point it there.
Pointing the omnis at what you want is usually the best answer. The high-frequency EQ fine-tuning aspect I mentioned is more of a subtle nuance thing, quite microphone and situation dependant. More importantly (and were Page and I begin to diverge) on-stage spaced omnis spaced at least 3-4' wide (and/or using more than two of them) leverage both proximity signal-level effects and time-of arrival effects. That can be useful in providing a nice balanced pickup of all on stage sources, even with players moving around, compared to a pair placed in one central spot. I find they provide far more imaging effects than just >4kHz, but they need to be spaced enough to do that. They'll still work and sound natural more closely spaced, but then they don't provide much imaging below the treble.
On stage is more of a "capture the sound" question than ambiance since there is no intended ambiance on stage. All of the sound that is occurring there is occurring for a reason (monitors, amps, instruments, etc). Any ambiance or sound image is largely unintentional except in a few genres.
Here's were we diverge a bit more strongly. What I get from on-stage which I can't get elsewhere is primarily excellent and detailed imaging and an on-stage ambient feel (talking about an on-stage stereo pair here, or multiple spaced omnis rather than close-placed or spot mics which only provide direct, SBD-like sound). Secondly, on-stage omnis provide both room and audience ambience as well. Unless I have a dedicated ambient pair facing out into the room or an AUD pair, all the ambient room information on the recording is being provided by those on-stage mics. Those aspects are completely intentional and one of the things I consider when arranging the on-stage setup.