From the two application circuits shown in the opening posting, only the first one can be recommended.
The 2nd one cannot go down to low frequencies unless L1 inductance is very high, which would make it unusable at high frequencies in turn.
Then the (unlucky) combination of L1, C3, C4 and C5 looks like a good idea, but usually isn’t in practice, because it forms a multi-resonant circuit that might cause all sorts of troubles at high frequencies, depending on the individual component properties and the mechanical construction of the circuit.
The first application circuit might provide slightly less gain, but has significantly wider bandwidth and is much more benign. The resistor Rbias has to be rather low to provide the required (high) bias current, hence it causes additional load at the amplifier output, which in turn will reduce the gain especially at high frequencies. This is why the optional RFC can be added, so that the amplifier only sees the actual load connected to the out terminal at high frequencies. The inductance should be optimized so that the gain is as flat as possible, which would be hard to predict in advance, because there are so many variables. It has to be tested in the actual circuit. For instance, the Resistor Rbias will have some inductance already, but also some capacitance, which makes it a parallel resonant circuit and its resonance frequency should be above the upper bandwidth limit of the amplifier. Yet it might be just right and that’s another reason why RFC is optional.