Series RLC circuits

In both instances, context makes it clear that you're describing an RLC circuit configured as an RLC bandpass filter, probably either like this or like this.

In the first case (one signal at a time), you note that the output is maximum when f_in = f_r (and, therefore, less than maximum for any signal f_s != f_r). This is totally reasonable.

In the second case (two signals at once), superposition dictates that you'll just get out of the sum of what you'd get for the two signals individually (something that castle-bravo's answer also states).

In a third instance (infinite signals / signal+noise), which you didn't mention, the same thing happens. The signal at f_r makes it through, while everything else is more attenuated.

Understand that you're still describing a filter. That's the net effect. If signal f_r passes through attenuated slightly and signal f_s is attenuated a lot, you end up with a filtered signal at frequency f_r.

Of course, one pass through the filter won't eliminate all the other frequencies, but it will certainly help, especially if f_r and f_s aren't close and the Q factor of the RLC circuit is high.

In theory, R, L, and C circuit components are linear, which means that if the circuit responds in a particular way due to one stimulus, and another way due to another, the response to a signal which is the sum of the stimuli will be the sum of the responses.

Put another way, in linear circuits where $I$ is a function of $V$, $I(V_1(t) + V_2(t)) = I(V_1(t)) + I(V_2(t))$.

To answer your question more directly, if you have a current $I_1$ due to signals of frequency $f_1$ and $I_2$ for signals of frequency $f_2$, then the resulting current from the sum of the two signals will be $I_1 + I_2$.

Things get more complicated when you consider high-frequency behaviour of circuit components, heating effects due to current, and other effects, but I think these are outside the scope of your question.