On 1., you may be under the misimpression(e.g. your reference to "power-hungry" speakers) that the type of biamping involved increases the amount of power available to the speakers. This isn't the case, since there's still only one power supply section and it's simply feeding available power through two channels of output transistors instead of one. Each section of the speaker would still have 120 watts available to it, not 240. Of course, although there's no significant benefit, neither is there harm in leaving it that way if you don't have a better use for those channels, e.g. driving back surround speakers.
On 2.and 3., manually balancing the levels by ear can still be considered "calibration"; only doing nothing at all would leave the system uncalibrated. The question would be what's easier and/or more accurate. My view would be that the answer is getting the 3805 mike and letting it do the job. A microphone is probably(although not necessarily, since the RS meter, for example is +/- 2dB)going to give a more accurate level adjustment than our ears will. Removing the human error element by feeding the mike data into an auto-calibrating receiver is again more likely to give an accurate result. So, letting the receiver do it is likely to be both easier and more accurate than a manual calibration either by ear or assisted by a separate SPL meter. Another factor is the ability to at least try the room equalization to see if the result is pleasing. Of course, after any of this, the result can be adjusted to suit your taste.
