I don't believe that the carrier still exists because of tank circuit ringing when 100% negative modulation is reached either. The Q of the tank only allows for a couple of cycles at most when no carrier is present. It decays very quickly. With modulation, even at voice frequencies, the modulation frequencies are much lower and there are many more rf cycles for each audio cycle. So one audio cycle would be equivalent to many rf.
It wouldn't be the ringing in the transmitter tank circuit that would fill in the "holes" in the carrier. As you say, the rf tank would only smooth out the rf waveform of the carrier. The bandwidth of the resonant tank circuit is tens of kHz. It is ringing in the i.f. filter in the receiver that produces the effect. The i.f. selectivity is ideally about the same as the bandwidth of the transmitted signal, and with today's receivers would rarely be more than twice the bandwidth of the signal. The receiver bandwidth must be many times the signal bandwidthl for it to appear as an "amplitude modulated" carrier rather than steady carrier plus sidebands.
Neither approach (physical existence of sidebands or carrier of varying amplitude) is "wrong"; they are completely campatible with each other. It is a matter of how you "look at" the signal. With a wideband receiver such as an untuned crystal set or cathode ray tube deflection plate, the existence or non-existence of sidebands is meaningless. With a selective receiver, the carrier of varying amplitude becomes an impossibility as the frequency of variation approaches the receiver bandwidthr.
This is somewhat analogous to another concept of physics; the quantum theory. Electromagnetic radiation such as rf and light have a dual nature in that they can be equally well thought of as particles (photons) or waves, but not simultaneously.
Don K4KYV
