There's a 4th you're missing: lower reciprocating mass.

Low mass connected to the wheels(aside from the stator) combined with near instant torque control means they can adjust power to the wheels in a sub-ms(supposedly) range. Once you break traction it's really hard to get it back and you're fighting the momentum of an ICE engine in the non-EV case. You also have two completely independent motors so you don't have to fight with any transfer case or torque converter and it can just instantly shift power to where traction is available.

All I know is I can floor our 85D in the rain and it doesn't even shimmy, just plants and takes off.

The fundamental explanation is that electric drivetrain's efficiency curve is rather flat, meaning that you can build a powerful drivetrain without sacrificing efficiency (= range in electric car).

With ICE, you have to choose either power or efficiency, i.e. the engine is tuned for specific load and the efficiency sacks when heavily under loaded. (Constant legal-speed driving makes only a fraction of load of heavy acceleration.)

The "cold starting" issue isn't as significant as it might seem. Most acceleration tests on ICE vehicles are started with the brakes down and the throttle up. It is very hard on the vehicle and doesn't match what you would likely do in the real world, but it does mean power is available right from the start.

It's not equal torque at all rpm.

Effectively it's more equal then an ICE.

Also if you can get someone who has a lot of experience with motor control they can get you going pretty quickly. I'm assuming what's in the Tesla is a brush-less DC motor and there are a few things you can do (especially since you're on some serious batteries in the Tesla) to push some serious torque at the startup curve. You're probably only running a modulated power start for a few seconds. After the first few seconds you're shoving the full amperage into the motor.

As pointed out in other comments Tesla actually uses AC induction machines (asynchronous AC) instead of brushless machines (synchronous AC). To a first approximation an AC induction machine can produce constant torque up to a fixed speed. Above this speed the torque falls off as approximately 1/speed.

You are right about the startup torque. AC induction machines can produce full breakdown torque at zero speed, but this does require high current (but luckily not high power because the applied voltage is still low due to low motor speed).

Another limit is the junction temperature of the power semiconductors in the inverter(s). I think Tesla uses liquid cooled inverters, but this typically means there isn't much mass to act as a heatsink (relying on the liquid coolant instead), so overload times are typically very short. I'd would guess this is actually the limit in their design. In other words they can put very high currents into the motor for a long period of time before it thermally overheats, but way before that time they have to reduce the current in order to protect the power semiconductors in the inverter.

Teslas have AC induction motors.

If they do that then they can also take advantage of mutli-phase power that they could modulate to change the speed thereby running it at a higher torque at lower speeds.

So, torque is defined as Tau = P / omega where P is power, omega is angular velocity and Tau is torque. In an ICE, P varies a lot with rpm (i.e. angular velocity) giving a sort of inverse-bathtub curve whereas in an electric motor the power curve is pretty much flat. In an ICE at startup (so essentially zero plus epsilon rpm or angular velocity) the power is quite low, so torque is also low. In the electric motor the power output will be essentially whatever it is rated at, so the torque will be highest at this point, getting lower as rpm (angular velocity) increases.

For a much better explanation than I can provide, see https://en.wikipedia.org/wiki/Power_band#Electric_Motors