I guess the real question is why does air above 70F lend itself so much better to lean-burn?
A _VERY_ complex question redbug. With several factors all contributing or detracting. Dougie's got one covered.
First off remember more power doesn't necessarily mean better MPG. You'd think that if the same engine can be coaxed to yield more power and you simply don't use it then it would consume less fuel since it doesn't have to "work" as hard. Which is true in part. This is where lean-burn comes in and turns that thinking upside down.
Warmer weather vaporizes fuel better allowing for more complete combustion. Its a basic principal of all (AFAIK) liquid fuels commonly used today. Which effectively rules out anything not petroleum based. Even the fuel stingy Insight _requires_ a richer (greater fuel to air ratio) on a cold (ambient temperature) start to compensate for the inefficiencies. Thats one factor, getting the car warmed-up. And the lower temp you begin from the longer (and more fuel) its gonna take to get it up to operating temp. About 195F (90C) coolant temperature is fully warm.
In _cold_ weather this effect is still evident and more pronounced in an Insight _because_ of its fuel stinginess. Many members have demonstrated that the Insight won't reach operating temperature at idle in _cold_ weather. There's simply not enough fuel being consumed by the car with enough waste heat to compensate for the losses. Part of the reason that auto stop is disabled in temps 40F and below.
Pre-warming the intake air is a system that has been used in cars for _many_ years now. The problem used to be carburetor ice. Ice will form under some conditions due to atmospheric moisture and the evaporative cooling effect of the fuel mist. As you can imagine a car with a plug of ice in the intake will have serious power and performance problems. MPI (Multiport Fuel Injection, 1 injector per cyl usually mounted in the intake manifold and pointed at the intake valve) is not affected by the "old" carb ice condition so adding heat is not commonly used. There is a certain amount of heating the intake air that occurs because of engine and its radiant heat losses and the resulting under hood temperature. In colder weather the under hood temps will be correspondingly lower further reducing intake air and fuel temps (heating of the fuel rail).
The dynamics of lean burn require a good understanding of the chemistry and physics of combustion. To oversimplify; a lean charge tends to burn more slowly. Given x RPM's the exhaust valve is _going_ to open up in a predetermined amount of time regardless of whether combustion is complete. So simply "forcing" a lean mixture past a certain and _highly_ variable point (load, RPM, temperature and several other factors) will fail to produce power.
With a hot air mod in optimum temperature conditions you can partially simulate summer to the engine. Warm the intake air to 110-120F in cool conditions and you'll see a "widening" of the lean burn window. The theory is that when things get cool there's simply not enough time for the fuel to be sufficiently warmed by the intake air to vaporize as effectively and lean-burn doesn't like it. What's happening in reality is still open to some speculation. Honda doesn't publish all the parameters of engine operation that the lean-burn system is designed to work within. There is one, a lean burn speed limit for the Insight, 72 MPH ish ( :?: ) The stratified charge concept is the basics of what allows lean burn and has been used by many Honda engines for almost 25 years. The concept is published in various sales brochures and technical manuals.
Now I know what your thinking.
Warm the fuel :!: Its done with diesel because in low temps it begins to congeal (turns to jello) and its not as _HIGHLY_ flammable as gasoline so its relatively safer to do. Ok then, let's warm the intake air even more. Well Rick Reece tried a 150F (HOT) intake air mod and during the coldest part of winter with no apparent MPG improvement. Probably related to your original thinking of colder air = more power. Heat the air too much and it thins too much, correspondingly reducing the available amount of O2 for combustion and resulting in a loss of power.
And then there are the additional factors of more aerodynamic load at speed with colder (more dense) air and less flex in the tires. And the different winter fuel blend and the... and... and...
Anyways there it is in a nutshell.
IIRC the coolant flow to the TB is there to normalize its temperature relative to the rest of the engine. Else fuel vapors traveling back up the intake will "freeze" there and more rapidly gunk up the TB. And the ice problem is still an issue. There is a temperature drop from the resulting pressure drop across the throttle plate. Although without the added evaporative cooling effect of fuel, like in a carburetor, I'd suspect ice is a rare occurence.