"Getting back to DeltaV: Why is a 4.0 Km/s deltaV to EML so great when it is 3.2 to reach the Moon and 2.7 to land (which I think you need even coming from EML) and 3.6 top reach Mars and 1.0 to land."

There are different ways of getting to Earth/Moon L-1 &/or 2. I don't know where this 4.0 km/sec number comes from- I would guess it's the delta V needed to sprial out of LEO using low thrust- such as with Ion rockets.
Any Earth L-point could require less delta-v to get to than compared to getting to a Low Lunar orbit. You can get to EML 1 or 2 for about 3.1 km/sec from LEO. Of course, you have to also remember it costs additional delta-v to first go to LEO, than compared to a GTO type trajectory or escape trajectory.
The european Smart satellite, is starting from a GTO type orbit and is sprialling out with ion type rockets. It's going to require an excessive amount of time to get to a lunar orbit [though less than going to Mars].
Taking this longer duration route to get to the Moon is in my opinion a wise move- because it's provide experience in how to operate in these lower delta-v routes of moving anything in space and allows them to use this newer type of propulsion.

If you consider that it takes "about" 9.0 km/sec to get to LEO, if instead you start with a more escape-like trajectory instead of LEO- the amount added to 9.0 could instead be around 2.0 km/sec- so instead of a total of 9 + 4 = 13 km/sec, you could have around 11 km/sec. Or if you take your 9 + 3.2 + 2.7 = 14.9 total to land on the Moon. A GTO type trajectory to Lunar surface could "around" 13 km/sec and adding in a stop to EML 2 could cost around another .5 km/sec to total delta-v needed.
Now in LEO you need to spend delta-v for station keeping, in EML you would need to spend less delta-a for station keeping- if you doing a manned Mars *non-direct* then you could easily have longer than 6 months worth of station keeping.

Here's a direct quote from ESA:
"Mission controllers must then inaugurate a new era of space navigation. For the very first time, they will use the sustained thrust of electrical propulsion jointly with manoeuvres under gravity. Isaac Newton knew nothing of such tricks, and ESA's experts have had to invent fresh mathematics for figuring out the best orbits.
Joseph-Louis Lagrange
The tug of the Moon's gravity will at first help to widen the spiral orbit, in regular encounters called 'lunar resonances'. By the time SMART-1 passes within 60 000 kilometres of the Moon, the effect of gravity will be much more pronounced, in encounters known as 'lunar swing-bys'.

At a crucial stage in the journey, called lunar capture, SMART-1 will pass through an invisible doorway in space at Lagrange Point No. 1, or L1 for short. As first noted by the mathematician Joseph-Louis Lagrange in 1772, the gravitational effects of the Moon and the Earth are in balance at L1, 50 000 to 60 000 kilometres out from the Moon on the earthward side. "
SMART-1