bpf: Add "_if" tap APIs

[jhibbits]

Summary

Hide more netstack by making the BPF_TAP macros real functions in the
netstack. "struct ifnet" is used in the header instead of "if_t" to
keep header pollution down.

[glebius]

Do we really need the "_if" suffix for the new functions? Why don't name them the same way the old macros are named, just lowercase them?

[jhibbits]

In D38103#867115, @glebius wrote:

Do we really need the "_if" suffix for the new functions? Why don't name them the same way the old macros are named, just lowercase them?

That would work for the ETHER_BPF_MTAP() analogue, but not the BPF_TAP(), BPF_MTAP() and BPF_MTAP2() macros, those lowercase names are already KPIs that take a BPF pointer. I named ether_bpf_mtap_if() as I did to keep parity with the bpf_tap/mtap_if() names. I considered just co-opting the bpf_tap(), etc names instead, but figured they're KPIs that are probably used by others and didn't want to step on toes there.