Based on your first-hand experience shooting black powder muskets:
1. What is the typical misfire rate of percussion cap small arms?
If it is clean they work fine. After a certain number of shots, the fouling in the barrel renders loading difficult. The US issued a Williams patent cleaner bullet in the package of ten cartridges, designed to blast fouling out of the bore when fired, and relieve that issue. Every package of ten cartridges also included 12 percussion caps to make up for dropped caps or misfires.
live-firing a reproduction Enfield rifle-musket, with modern .58 minie type balls (very closely fitting in the .58 bore of the replica) on one occasion in the late 1990s, I found ramming became difficult after about 20 rounds. No problem with ignition.
2. What is the typical misfire rate of flintlock small arms?
When clean, and the flint fitted correctly, they work fine. A well knapped gun flint was generally good for about 50 shots, if well and securely fitted in the jaws of the hammer. One flint to 20 cartridges was apparently the common issue.
Firing a replica Charleville musket, I've found that maybe 25 shots with a perfectly tuned up and clean musket can be achieved (the US cartridge boxes for flintlocks generally had provision for 26 rounds in the wooden case, with a reserve below in the tin.
Things start to go wrong fast. The ball was sub-calibered (.64 inch for the .69 inch bore) so loading despite the fouling in the bore was not a big problem. But the fouling can jam the vent hole at the breech, and at least with replicas and modern powder, in some conditions the cake of fouling on the underside of the flint or frizzen can flake off onto the powder in the pan, causing misfire if not wiped off between shots (R.S. Dorsey claims in his gun tools books, that modern powder fouling is distinctly different in a manner than mid-1800s, so maybe that was not as bad back then). The flint can break or loosen in the jaws or fall out, etc.
Friends who shoot original flintlock muskets have demonstrated they are less liable to misfire than the replicas for reasons beyond the point of this thread.
The British in 1834 estimated perhaps all told an average of one misfire in 6.5 shots for a flintlock, and one in 165 shots for percussion from their experiments: