I don't see why you shouldn't be able to use the config. in #4, but with a circuit that simple, I think I'd start by building it up on some protoboard and see if it works. It's only like, 7 parts or so. Maybe you could add a choke to the voltage supply on the op amp.
Not sure what you mean by protoboard, but at this frequency I can't recommend regular stripboard/breadboard techniques.
To the OP, there are plenty of designs but most use tuned and matched discrete transistor stages or brick modules (both offer out of band filtering) or MMICs rather than op amps. Taming high speed opamps, even those a decade or two slower than the one you cite, is an art in itself.
I also don't know what your local regs are, but generally the band is shared globally and secondary/ISM users must accept interference from primary users. Bear in mind those primary users may well be applications with military, radar or space use, and they won't take very kindly to interference from unlicensed or secondary users.
The third harmonic is often used by aviation radionavigation, and although they model out interferers it won't take much for your signal to appear, and bearing in mind the perceived safety-of-life aspect you need to make sure that whatever you come up with isn't going to upset someone.
You should check your local regs if you haven't already.
The network on the output is trying to match the impedance of the transistor output stage to the antenna to obtain maximum power transfer. Remember maximum power transfer theorem at DC? This is the same, but at AC. It will also have some filtering effect, but not a whole lot judging from the circuit. You must bear in mind that at this frequency parasitic inductive and capacitive components come into play so layout and construction techniques become critical. With the schematic as shown, it's as much about what's shown as what isn't.
To match one complex impedance to another, it's common to use a combination of an inductor and a capacitor. There are online calculators to do this, but in the old days we used Smith Charts.
http://www.daycounter.com/Calculators/L-Matching-Network-Calculator.phtmlhttps://home.sandiego.edu/~ekim/e194rfs01/jwmatcher/matcher2.html+ many others.