I don't suggest trying to re-record fill. It won't sound the same for a variety of reasons. It's best to dig through your edits and find any and all small tidbits of fill for each "noise profile" and create long fill tracks (the Reverse tool helps a lot with this). Once you have a good 5+ seconds of steady fill that's free-and-clear of sound oddities and pitch fluctuations, here's what I suggest that may be a shocker:
Play both fill (angle edits) at the same time continuously.
It's a "least common denominator" situation.
Why do this? You mask the perceived noise bumps between the two types of noises - just run the edited track and their backfill all at the same time. By adding noise you will psycho acoustically "remove" perceived noise. I have had great success with this in even the most facepalm of dialogue scenes. In some cases when you stay with one mic/angle for a very long time (15+ sec) you can gradually get out of one of the noises... but if it's rapid-fire dialogue there's no such opportunity without drawing attention to it.
Then when you're in the premix stage, the common noise that is tied together can be processed with a real-time noise suppressor and the levels can be sculpted around the lines on the premix master fader (technically an aux track) - rather than wasting time trying to cajole three different noise profiles to work together in a back-and-forth crossfade dance. Sculpting the levels is helpful in very noisy circumstances, yet must be done careful so as not the pump the dialogue noise suppressor or create an artificially unattractive compression pumping quality where noise hisses in-and-out. This sculpting of levels is a careful balance which takes time and experience. There's a particular scene in the new Robin Hood film when Russell Crowe is down a by river in daytime and taking with somebody else - it is an great example of this 'hiss pumping' and what can happen when done wrong. Maybe it was fixed for the DVD/BluRay mix yet in the theater it was very unnerving to my ears.
Plus, this is a driving scene. By creating a steady, albeit slightly louder, noisefloor by running both noises, at least it will be steady in pitch and level - steady enough that your cut sound effects of an engine interior drive will mask most of it's nastiness (plus any wind buffet/whistle you have in the car too). If you had noise bump in pitch and level by not running them at the same time and just going back and forth, such masking would not work. Also, if the best noise you can find for backfill has a little bith of seat cloth or rattle movement, your suspension and seat creak sound effects edits should mask that or blend with it. It's not ideal to have this in the backfill to begin with yet I'm considering a worst-case dialogue scenario here: being in the mind of the sound effects editor and how they can help you.
This is only my personal dialogue editorial opinion and experience, based upon dealing with both crystal clear and atrociously gnarly sound situations that share this multi-mic or multi-angle complexity. It is by no means a be-all-end-all. It may just be a good place to start with and see where it goes.