This is difficult to answer, since you seem to have a major misconception about forces. The best answer is to go back your freshman, or even high school, physics book and read the section on forces. However, briefly:
The pulling force isn't somehow split between the train and the ground. The rope will pull with the same force on whatever is holding each end. If the rope is pulling on the train with 300 N, then the man is pulling on the rope with 300 N, and the man must be pushing on the ground with a force of 300 N laterally. The vertical component of the force the man is pushing on the ground with will be his weight plus some fraction of the pulling force depending on the angle of the rope.
If the rope is horizontal, then no part of the pulling force results in vertical force on the ground. In this case, which is pretty close to your bottom picture, the horizontal component of the force on the ground is equal and opposite of the force the rope is being pulled with, and the vertical component is simply the man's weight.
The question is quite complex, but there are several fairly simple things that can be observed:
First off, the standard bicycle, as a machine, is quite efficient. Very little energy is lost in the "drive train", with the vast majority of "lost" energy (mechanical energy input at the crank that is not converted to momentum) being expended as either air friction or friction between tire and roadway.
The human body, on the other hand, is often an incredibly inefficient machine. Not only are there simple concerns of "energy efficiency" -- how many calories of food, say, it takes to produce an erg of "work" -- but there are also major issues of "durability" and "endurance", both in the short term and long-term.
The average human body tends to have a "sweet spot" for cycling where the cadence is (depending on the individual and the circumstances) somewhere between maybe 60 and 90 RPM. Cycling within the "sweet" range for the individual produces a large amount of energy (though perhaps not the "peak" energy) and, more importantly, minimizes fatigue and optimizes endurance (as measured, say, in total energy produced in a given 24-hour period, including rest, eating, sleeping, etc).
In terms of gear ratio, in addition to determining cadence on relatively level ground, it also, of course, affects climbing. An individual is limited as to the total torque they can place on the bike crank arms, and hence what degree of incline they can climb at a given gear ratio. Lowering gear ratio (obviously) reduces the torque required to turn the crank arms and hence enables climbing a steeper incline. Here the "sweet spot" (for a relatively short climb) is below the level ground "sweet spot", but there still is one.
When considering cadence both on level ground and climbing it needs to be considered that muscles are more efficient when in "aerobic" mode -- burning "fuel" using oxygen supplied from the lungs via the bloodstream. Aerobic mode is perhaps twice as efficient as anaerobic mode (though don't quote me on that number), and, of major importance, it produces far fewer metabolic byproducts which can accumulate in the body and eventually become toxic. Although there are several factors that determine whether exercise is aerobic or anaerobic, a major one is, in fact, cadence, with lower cadences being more likely to be anaerobic.
Best Answer
One word: inertia. When you're riding a bike on a level gradient you just need to give it a push to get going, then you can coast for quite a while before friction and air resistance slow you down. In other words, the relatively frictionless wheels mean the bicycle's kinetic energy doesn't dissipate quickly. But the human body doesn't have wheels, so while running you have to give a good kick to get going, and then another kick to keep going on the next step, and so on. When hills are involved the difference is even more pronounced, since we run downhill the same way we do on the level, by continually pushing ourselves forward; whereas on a bicycle you can take advantage of the slope and just coast down it.
I suspect that raising and lowering your centre of mass isn't as inefficient as the other answers have suggested. This is because your legs are springy, so at least to some extent you're just converting energy back and forth between gravitational potential and the spring force in your legs. Humans are possibly the most efficient long-distance runners in the animal kingdom. There is a school of thought that says the reason we are bipeds is that we evolved as endurance hunters, chasing our prey until it collapsed from exhaustion rather than trying to outrun it over short distances. Whether that's true or not, we probably wouldn't do all that bouncing up and down if there wasn't a good reason for it.
You might ask why, if using wheels is so much more efficient, didn't we evolve that instead? I don't know, but it seems no animal has been able to evolve wheeled locomotion.