1. Any EM wave.
2. High-reflection dielectric stacks (which essentially behave as a whole bunch of parallel mirrors stacked on top of one another) are often called 1-dimensional photonic crystals because they exhibit the photonic band-gap effect in a single direction.
http://en.wikipedia.org/wiki/Photonic_crystal
To answer you question - Photon's cannot be generated within the "cavity" as there is technically no optical mode available to radiate into, and photons cannot be injected into the "cavity" as they would be reflected, again, due to no optical mode being available within the "cavity" - this is the essence of the so-called photonic band-gap effect. Essentially, there is no way the energy could have gotten where it is in the 1st place - your initial proposition is flawed.
For questions 3 & 4, it is more correct in these instances to refer to the phase of a photons
wavefunction rather than the phase of a photon.
3. Here you want to speak in terms of photons/sec, to establish and maintain an EM wave of any amplitude, you need a constant stream of photons, since the wave is constantly radiating energy. Photon flux (photons/sec/m^2) is proportional to
irradiance (in W/m^2). For a coherent wave, the irradiance is proportional to the amplitude squared, hence increasing the photon flux by a factor of 100 results in a 10x increase in amplitude.
4. No, you can't. Again, it makes no sense to "add photons in perfectly out of phase" it only makes sense to add two photon wavefunctions that are perfectly out of phase. When you do this, the result is obvious, there is no EM wave, no photons, and thus your philosophical proposition becomes redundant. Note though there is no implication that there are photons zipping in and out of existence, as is the implication if one were to refer to the phase of a photon rather than a photons wavefunction.
Claude.