And reinforcing b7 at the same time.

There really is little for white here now. Perhaps better was either Be3 or Bd2. Trading bishops, while it heads for an endgame when down a pawn, would remove a potent attacker disrupting the white kingside. It buys time. Black would then need to bring other pieces into play to continue the attack.

White did gain time here though. With the Bishop removed, none of Black's other pieces are positioned to quickly support and continue the attack. The knight can continue the attack, but, after repositioning the queen, g4 expels the knight.

But if white plans a kingside assault --it seems his only option-- was Qf2 best?

What complications? If white tries g5, it creates a knight outpost on f5 only second to e4. Not only would this outpost be strong, it would undermine white's kingside push. The other pawn advance f5 is premature and allows Nxg4. Black has time to prepare a defense to kingside expansion. It would seem that black lost time here because the knight threat against the queen, when it was at e3, was premature. Now he retreats.

Correction: Nh6 not Na6.

I prefer f5. The knight is adaquately gaurded and if ...exf then gxf followed by a quick f6. No matter how black responds, white will own g7. I believe this idea could change the balance of the game. White no longer has time to be too cautious, but a failed because of a lack of preparation is still a failure.

White loses time defending the knight, but f5 fails without it.

Preparing Rc1.

White shows he isn't following planning f5. Re1 now seems premature.

Once Rcc2 happens, white's kingside initiative seems doomed to fail. Rh2 seems called for because white's strongest piece is tied to the defense of a minor piece.

I believe that either Qh4 is superior. Good moves played without preparation will fail too. Black will eventually force Rh2, but, with Qh4 followed by h6, white can put enough pressure to almost force the exchange of rook for knight. The idea of f5 is probably too slow now.