CO-05 is a bit of a shocker, and vaults that one into Leans GOP. If nothing else, this might mean that a seat flips that wasn't on my list as of Labor Day, which would doubly disappoint me. :(
Meanwhile, the CO-04 poll presents something of a dilemma. In the Survey USA poll that showed a 4 point race the Reform Party candidate was named and took 8% of the vote. It's unclear whether Mason-Dixon similarly named the third party opponent or which approach is preferable anyhow. However, Colorado voters have a history of giving third parties a sizable protest vote commonly in high single digits.
About CO-04: Mason-Dixon polled a generic Other candidate @ 7%, which I'm sure you noticed. Usually, in polling, when refering to a generic Other candidate, you get a higher % than when you actually poll the name of the candidate unless that person has significantly higher name recognition.